91.. ”V4.44. .. 4» KW“ WWW...) . may .. .....34. ....45 v.11». . . A a. “9—..- .....24...‘ . . a. .o’ ....n.....w,./ Away.” 1.. 4 «C...— 4.......... .... ”Wt/$4.104... ...r or I» 4.. ..lnfa ... .1145"ng .... .....41... a.......4:.. .... . ... ..- ... 7: D V/ufllqM. .. Hurlfl 114.4 7 AMWWN.” 9.74. z Mfiflfiv.” . ”.../u 2;“. .flm ”I: ... Mr aflm nanny 94...”. .u» WWW. law-”... 4.54..u 4.14...” PE ANN .. fl- :Pfi ".7 .. ..4. . . 5.4 ......h... 4. 4. . f KGROUND: 50 am ,6: jivi? #51:}: 'L :2". {e .144: .flrnvifi. vim/447.... 073.04.“? 41" . . ...WWHW} ... .mlmmrmme a. Wm... ,4 .../,4.» 1.111. c .4 a. 14. «WM. 4 1.1.7....” FUN c. BAG Ml "WW 0?: AN NT 35?. 0'02 E R TU 0 STA SHH’B . 0 0. _ . . . .m n. . C .. .3 H: ”m.mw H .hm,mW. w M a . ... T . .. . .. ....vvvaonwvav THE 3000 . ACADEMIC. . . .. . 4 . . . ... .. .... 1.4... .... .. . ............I.4...... . .u... .....4...s¢..ao:»....».r. .431...- . . . .. ... . .. ... ... . ...... I... ...; ...... .....J... 6.5%.: 1.4.1.4.]...4u ...}.n.....4.f.m....ns.r...k..a.... -42. ...WPi. . -0 '.t I ' III I. I WNWi\liW“iiiimilwWNW i" 1.. qzfl“ . . .E University in ‘Rrw- 3. This is to certify that the thesis entitled The Relationship between Students' Socio- Economic Background and their Academic Achievement at Sixth Grade In Turkey presented by ALI DOGAN ARSEVEN has been accepted towards fulfillment of the requirements for ‘ Ph. D degree in Education // a; (1 {(0% Major professor 4 Date/7.4m; 23, {73 0-7639 ABSTRACT THE RELATIONSHIP BETWEEN STUDENTS' SOCIO-ECONOMIC BACKGROUND AND THEIR ACADEMIC ACHIEVEMENT AT SIXTH GRADE IN TURKEY BY Ali Dogan Arseven The Purpose The purpose of this study is to investigate some of the nonintellectual factors, namely socio-economic (SES) and socio-psychological factors (SP8), and their relationships with the academic achievement of sixth grade children in Ankara, Turkey, during the 1971-1972 school year. This study also attempts to compare students from a primarily low socio-economic population with students from a primarily high socio-economic population with respect to the relation- ships specified above. The major research questions explored in the study are as follows: 1. What is the magnitude of the relationship, if any, between a student's academic achievement and his socio-economic status? 2. What is the magnitude of the relationship, if any, between student's academic achievement and socio- psychological variables? Ali Dogan Arseven 3. Is SES or SPS more significant in establishing these relationships? 4. To what extent, if any, do SES and SP8 variables differ, between the two student populations, in predicting students' achievement in selected subjects? The Methodology The population under investigation consisted of two stratified student populations (primarily low SES and pri— marily high SES) attending sixth grade of seven public middle schools in Anakra, Turkey, in the 1971-1972 school year. A majority of the students of low SES live in Gecekondu dwellings (slum area) and the students in high SES live mostly in well-to-do neighborhoods (non—Gecekondu) in the metropol- itan area. The sample included 364 students from the low SES population and 378 students from the high SES pOpulation. Both samples were randomly and proportionally selected from their respective entire populations in this study. Two main sources were used to collect data for the study. Students' grades on reading, mathematics, and G.P.A. of five subjects (dependent variables) were obtained from school records. Information about students' socio-economic status and socio—psychological factors was obtained by means of a "Student Questionnaire," which was supplemented with a "Parent Questionnaire." Parents' responses to the items in the Parents' Questionnaire were used only to check Ali Dogan Arseven whether there was a consistency between students' answers to the answers to similar items in the Student Questionnaire. Student's educational background, father's occupa- tion, father's income, father's education, and student's residence condition were used as indicators of his socio- economic status. School aspiration, self-concept of ability, perception of the expectations of significant others (par- ents, teacher, and friend) concerning student's potentiali— ties with respect to academic achievement were used as indicators of his SP8. The data obtained for the study were analyzed through the use of descriptive summaries of item responses, in terms of frequency counts and percentages. Selected further anal- yses of data were conducted using statistical techniques of correlational analysis, factor analysis, multiple regres- sion analysis, and stepwise regression analysis. Major Findings of the Study The following major findings emerged: 1. For the combined population, there is significant relationship between students' socio-economic status and their academic achievement. The highest relevant SES fac- tors to achievement are father's occupation and father's education. 2. In comparing the two sub-populations, the relation- ships between SES and achievement are substantial for non-Gecekondu students, whereas those relationships Ali Dogan Arseven for Gecekondu students are either negligible or nonsig- nificant. 3. The relationships between SPS variables and academic achievement based on the combined population are significant. However, the magnitude of relationships is higher in the non- Gecekondu population than in the Gecekondu population. Stu- dents' perceived evaluation by parents and students' self- concept of ability are the socio—psychological factors contributing most to the variance in academic achievement of students in both pOpulations. 4. SP8 variables were found to be more significant than SES variables in establishing the relationships between achievement and the above nonintellectual variables (SP8 and SES). 5. For the Gecekondu population, a negative relationship was found between SES and SP8, while it was positive for non- Gecekondu. There was no correlation between SES and stu- dent's grade for the Gecekondu pOpulation. THE RELATIONSHIP BETWEEN STUDENTS' SOCIOHECONOMIC BACKGROUND AND THEIR ACADEMIC ACHIEVEMENT AT SIXTH GRADE IN TURKEY BY Ali Dogan Arseven A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY College of Education 1973 “t ACKNOWLEDGMENTS The author is deeply grateful for the personal guidance, support, and encouragement in the writing of this dissertation provided by Dr. Harry L. Case, his major pro- fessor, and by Dr. Maryellen McSweeney, who—~through her generous assistance--acted in an unofficial co-director capacity. The writer expresSes thanks to Dr. Wilbur B. Brookover and Dr. Cole S. Brembeck for the encouragement and assistance they provided during the coursework and disserta- tion stages of his doctoral program. The author wishes to express his appreciation to Mr. Nusret Karcioglu, Undersecretary, Ministry of Education, Government of Turkey; and to Mr. Sudi Bulbul, Assistant to the Undersecretary, for their efforts in getting permission from the Ministry in order to permit the author to complete his dissertation at Michigan State University. The author also expresses thanks to Dr. Ben A. Bohnhorst, member of the Michigan State University field team in Ankara, and the personnel of Planning, Research and Coordi- nation Office of Ministry of Education of Turkey. The writer is very grateful to his wife, Sabahat, for the love and encouragement she has provided, and the patience she has shown to the writer during the completion of ii his doctoral program. This measure, to her assistance. Finally, the author Cigdem and his son Suat for years of graduate study. accomplishment is due, in large is grateful to his daughter their understanding during his iii TABLE OF CONTENTS LIST OF TABLES . . . . . . . . . . . . . . . . . . . LIST OF APPENDICES O O O O O O O O O O O O O O O O 0 Chapter I. II. III. INTRODUCTION 0 O O O O C O O O C O C O O O 0 Statement of the Problem . . . . . . . . . The Purpose of the Study . . . . . . . . . The Significance of the Study. . . . . . . Limitations of the Study . . . . . . . . . Assumptions Upon Which the Study IS Based O C C O O O O O O O O O O O 0 Definition of Terms. . . . . . . . . . . . overView O ‘ C I C I O O O O O O O O O O O 0 REVIEW OF LITERATURE . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . An Overview of the Debate on the Issue of Environment Versus Heredity Within the Context of Human Learning. . . . . . . . Introduction to Related Studies. . . . . . Review of Related Research Findings. . . . Socio-Economic Factors (SES) . . . . . . . General SES Factors and Achievement. . . . Socio-Psychological Factors (SPS). . . . . Self-Concept of Ability and Achievement. . Summary. . . . . . . . . . . . . . . . . . RESEARCH METHODOLOGY . . . . . . . . . . . . Definition of the Population . . . . . . . The Sample . . . . . . . . . . . . . . . . Sources of the Data. . . . . . . . . . . . Description of the Instrument. . . . . . . Collection of the Data . . . . . . . . . . Statistical Treatment of the Data. . . Summary. . . . . . . . . . . . . . . . . . iv Page vi ix |—l OxlU'IH 11 14 15 15 15 19 21 22 28 29 34 37 39 39 40 41 41 43 45 48 Page IV. ANALYSIS OF THE DATA . . . . . . . . . . . . SO Characteristics of Respondents . . . . . . 50 Deve10ping Indices on Selected Items and Reliability Analysis . . . . . . . . 60 Correlation Analysis . . . . . . . . . . . 63 Factor Analysis for Developing SES and SP8 Indices. . . . . . . . . . . 75 Multiple Regression Analysis . . . . . . . 85 Stepwise Regression Analysis . . . . . . . 99 V. SUMMARY, FINDINGS, AND RECOMMENDATIONS . . . 114 smary I I C O C U C O O O O O O O O O O O 114 Findings Of th Study and Discussion . . . 116 Recommendations for Planning and Research. 130 BIBLIOGMPHY O O O O O O O O O O O O O O O O O O O O 134 APPENDICES O O O O O O C O O O O O O 0 O O O O O O O 140 10. ll. 12. 13. LIST OF TABLES Distribution of Subjects by School in the Survey . . . . . . . . . . . . . . Distribution of Subjects by Sex and by School Category. . . . . . . . . . . . Distribution of Subjects by Age and by School Category. . . . . . . . . . . . Distribution of Subject as First Year Attender or Repeater in Sixth Grade, by School Category. . . . . . . . . . . . Distribution of Subjects by Their Father's Occupation and School Category . Distribution of Subjects by Their Father's Income and School Category . . . Distribution of Subjects by Their Father's Highest Educational Level and by School Category. . . . . . . . . . Distribution of Subjects by Existing Home Facilities and School Category. . . . . . Distribution of Students by Their School Aspiration and by School Category . . . . Means and Standard Deviations of Socio- Psychological (SPS) Variables by School Category . . . . . . . . . . . Internal Consistency of Indices Based on Selected Items . . . . . . . . . . . . Intercorrelation Among SES and Achievement Variables of Combined Sample. . . . . . . Intercorrelation Among SES and Achievement Variables of non-Gecekondu vs. Gecekondu sample. 0 O O C O O O O O O O O O O O I 0 vi Page 45 51 52 52 54 54 56 57 58 59 64 66 68 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. Page Intercorrelation Among SP8 and Achievement Variables of Combined Sample. . . . . . . . . 72 Intercorrelation Among SP8 and Achievement Variables of non-Gecekondu vs. Gecekondu sample. I O O O O O O O O O O O O I O O O O O 73 Intercorrelation Matrix of Selected SES and SP8 variables 0 O O O I O O O O O O O 79 Factor Loading Matrix of Unrotated Factors With SES and SP8 Variables. . . . . . . . . . 81 Rotated Factor Loadings on Two Orthogonal Factors. . . . . . . . . . . . . . 83 Intercorrelations of the Predictors and Dependent Variables Based on Combined Sample. 87 Beta Weights and Multiple Correlation for the SP8 and SES Indices in Estimation of Academic Achievement. . . . . . . . . . . . . 89 Intercorrelation Among Selected Variables (non-Gecekondu vs. Gecekondu) . . . . . . . . 9O ANOVA Table for Testing the Equality of Regression Equation with SP8 and SES Indices (Independent Variables) for Predicting READING on Two Samples at .01 Level . . . . . 93 Testing the Differences of Beta Weights Across the Sample for Each Independent Variable (SP8 and SES) With READING at .01 Level . . . . . . . . . . . . . . . . . . 95 Testing the Equality of Regression Equation With SP8 and SES indices (Independent Variable) for Predicting MATHEMATICS on Two Samples at .01 Level. . . . . . . . . . . . . 96 Testing the Differences of Beta Weights Across the Samples for Each Independent Variable (SP8 and SES) With MATHEMATICS at .01 Level. . . . . . . . . . . . . . . . . 96 Testing the Equality of Regression Equation With SP8 and SES Indices (Independent Variables) for Predicting G.P.A. on Two Samples at .Ol Level. . . . . . . . . . . . . 97 vii 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. Page Test of Significance of Differences of Beta Weights Across the Samples (non-Gecekondu vs. Gecekondu) at .01 Level . . . . . . . . . 98 Stepwise Regression Analysis for Dependent Variable of READING (non-Gecekondu Sample) . . . . . . . . . . . . . . . . . . . l02 Stepwise Regression Analysis for Dependent Variable of READING (Gecekondu Sample). . . . 103 Stepwise Regression Analysis for Dependent Variable of MATHEMATICS (non-Gecekondu Sample) . . . . . . . . . . . . . . . . . . . 104 Stepwise Regression Analysis for Dependent Variable of MATHEMATICS (Gecekondu Sample). . 104 Stepwise Regression Analysis for Dependent Variable of G.P.A. of Five Subjects (non- Gecekondu Sample) 0 O O O O O O O O O O O O O 105 Stepwise Regression Analysis for Dependent Variable of G.P.A. of Five Subjects (Gecekondu Sample). . . . . . . . . . . . . . 106 Stepwise Regression Analysis for Dependent Variable of READING (non-Gecekondu Sample). . 108 Stepwise Regression Analysis for Dependent Variable of READING (Gecekondu Sample). . . . 109 Stepwise Regression Analysis for Dependent Variable of MATHEMATICS (non-Gecekondu sample) a o o o o o o o o o o o o o o o o o o 11.0 Stepwise Regression Analysis for Dependent Variable of G.P.A. of Five Subjects (non- Gecekondu Sample) . . . . . . . . . . . . . . lll Stepwise Regression Analysis for Dependent Variable of G.P.A. of Five Subjects (Gecekondu Sample). . . . . . . . . . . . . . lll viii Appendix A. B. C. LIST OF APPENDICES Page STUDENT QUESTIONNAIRE AND PARENT QUESTIONNAIRE . . . . . . . . . . . . 141 CODING SHEETS. . . . . . . . . . . . . . . . 154 INTERCORRELATION TABLES. . . . . . . . . . . 161 ix CHAPTER I INTRODUCTION Statement of the Problem The trend in recent research studies on academic achievement appears to be toward investigating nonintellectual variables as important factors in explaining differences in academic achievement. Significant findings have indicated that intellectual measures account for only 35 to 45 per cent of the variation in academic performance.1 In recent education literature, there seems to be a growing concern, not only about sociological variables influ- encing learning and teaching, but also about the interaction among all of these variables.2 Mitchell pointed out that the determinants of learn- ing behavior need to be sought more often in the character- istics of the environmental context and the interaction of these characteristics with individual traits and abilities.3 1Mary Elizabeth McClelland, "An Investigation of Selected Non-Intellectual Variables and Their Relationship to College Academic Achievement" (unpublished Doctoral dis- sertation, Michigan State University, 1969). 2J. R. Campbell and C. W. Barness, "Interaction Analysis--A Breakthrough?" Phi Delta Kappan, L, 10 (June, 1909), 587-590. 3James V. Mitchell, Jr., "Education's Challenge to Psychology: The Prediction of Behavior From Person-Environment Interactions," Review of Educational Research, XXXIX, 5 (1969), 696. In certain instances, it seems that social forces and envi- ronmental contexts may be prepotent over individual traits, or they may have such immense implications and impact on individual behavior that they can not be ignored. Quite often, researchers who have focused on the indi— vidual learner appear unable to account for the variations in learning that may be due to the social environment. There- fore, it is necessary to come to a fresh understanding of the individual's environment and the relationship between his performance and his environment. In education and social science, researchers have recognized cultural and subcultural influences on the process of learning and social change, and are now trying to develOp adequate research designs to study the differences in learn- ing that seem to result from the differences in environment. An individual is born into a family that is part of a socially ranked group, and the family's social participa- tion generally is limited to that group. The individual's Opportunities for social mobility are limited by the pressure he receives from groups above him. It has been found that social classes Operate essentially to maintain barriers against intimate social participation with other social classes. For example, people of the slums are barred from intimate social participation with people from the middle and upper middle classes. A child cannot learn his mores, social drives, and values solely from books. He learns a particular culture and moral system from those peOple who exhibit this behavior, and who exhibit it in frequent relationships with him. If a child associates intimately with no one but slum children and slum adults, then he will learn primarily slum culture. Thus, the pivotal meaning of social class to the student of behavioral science is that social class limits and patterns the learning environment of the child. Davis pointed out that social classes form the structure of the social "maze" in which the child learns his habits and meanings.4 The child's social learning first takes place in the environment of his family and of his own play-group. The demands of a family upon the child differ significantly between the lower and the upper class. Usually, the child learns the values and beliefs that his parents hold. When the child starts going to school, his surroundings are broadened. He meets new people, and learns new values and habits through interaction with others in and out of school. Next to the family, the school is the most important insti- tution in which a child's socialization takes place. How- ever, elementary schooling does not have as much impact on the child's socialization as secondary schooling. The first adult community with which a student establishes close rela— tionships and in which he becomes an active participant is the secondary school. The secondary school classroom can be described as a living experiment for understanding the 4Allison Davis, Social-Class Influences Upon Learning (Cambridge, Mass.: Harvard University Press, 1960). problem of human relations in the larger and more diversi- fied social structure of a society. Classroom experiences provide the basis for a real conceptualization of a social system. The following theoretical concepts, which are widely accepted by social psychologists, provide a common basis for further analysis of the child's socialization. Brookover and his associates pointed out that typically human behavior emerges only from an individual's interaction with other per— sons who are significant to him within his environment.5 To understand the educational process and academic achievement of students in any kind of society, the social environment in which learning occurs must be known. The social environ- ment of any student somehow influences the knowledge, values, and behaviors which he acquires. The research of social scientists has brought us to the stage at which the concept of fixed intelligence is no longer functional. It can be said that heredity probably does set a fixed upper limit on intelligence. However, most students do not Operate near their maximum potential because of the limitations existing in their environment. There- fore, environment determines the extent to which an individ- ual approaches his maximum potential. Brookover and his associates pointed out that: 5Wilbur B. Brookover and David Gottlieb, A Sociology of Education (New York: Van Nostrand Reinhold Company, 1964), p. 16. Intelligence and other related aptitude measures are sample measures of what the individual has learned and do not measure directly any fixed or inherited capacrty or ability. The assumption of fixed ability still continues to dominate the practice and organization of education in many countries. Jensen7 hypothesized that there are inherited differences among individuals, and that variations in abil- ity cannot be explained through social class or environmental differences. Rather, the variations in intelligence must be attributed partially to genetic differences. Since the debate on the issue has not been concluded, the facts and evidences that have been found on either side are not reviewed here. Rather, the present study is primar- ily concerned with the environmental factors which have been assumed to have an impact on students' academic achievement, with special reference to sixth grade students in Turkey. The Purpose of the Study The purpose of the study is to investigate some of the nonintellectual factors, namely socio-economic and socio- psychological factors, and their effect on academic achieve- ment of sixth grade children in Ankara, Turkey, during the 1971-1972 school year. 6Wilbur B. Brookover and Edsel L. Erickson, Society, Schgpls and Learning (Boston: Allyn and Bacon, Inc., 1969), p. 5. 7Arthur R. Jensen, "Environment, Heredity and Intel- ligence," Harvard Educational Review, Reprint Series No. 2, 1969. The study also attempts to compare two types of student pOpulations, defined as low socio-economic and high socio-economic groups, by means of two sets of nonintellectual variables and their relationship to students' academic achieve- ment. The first set is categorized as socio-economic (SES) variables, which include father's occupation, father's income, father's education, student's residence, and his previous educational background. The second set is composed of socio- psychological (SPS) variables, which include school aspiration, self-concept of ability, and a student's perception of how significant others evaluate his academic ability. The data sought pertain to the following questions: A. Questions for combined populations 1. What is the magnitude of the relationship, if any, between a student's academic achievement and his socio-economic status? 2. What is the magnitude of the relationship, if any, between a student's academic achievement and socio-psychological variables? 3. Is SES or SPS more significant in establishing these relationships? B. Questions for comparing two sub-populations 4. To what extent, if any, do SES and SP8 variables differ, between the two student populations, in predicting students' achievement? a. in reading b. in mathematics c. in G.P.A. of five subjects——reading, mathe- matics, social science, natural science, and foreign language? The Significance of the Studyy Turkey is in the initial stage of its economic, industrial, and social development, and along with this develOpmental process have come numerous problems. One such problem is that since 1950, Turkey has been faced with rapid urbanization. In the last two deCades, many poor villagers from less developed parts of the country have moved into big cities, and these metropolitan areas have been surrounded with a kind of mushroom housing called "Gecekondu." According to Tutengil, there were 240,000 Gecekondu in Turkey by 1960.8 The percentages of Gecekondu dwellers with respect to the total population of Istanbul and Ankara were 21 and 45, respectively.9 Since 1962, approximately 170,000 peasants have been moving to the city annually.10 The construction of Gecekondu houses is very primi- tive. Research carried out in Istanbul in 1966 reported that each Gecekondu house has either one or two rooms, and on the average 4.8 people live in it. The head of the household is 8Cavit O. Tutengil, Az Gelismis Ulkelerin Toplumsal Yapisi (Istanbul, 1966), p. 101. 9Ibid., p. 102. loIbid. generally an unskilled factory worker who earns a wage of between two and four dollars per week.11 Urbanization is not a simple agglomeration of people in cities. Rather, it can be viewed as a complex process of social change. When one looks at the social aspect of Gece- kondu life, it seems that the people in Gecekondu dwellings, especially the older generations, still keep their rural culture within the larger metrOpolitan culture.12 What does one find concerning the schooling of chil- dren who were born and raised in that "Gecekondu culture"? One need not be a fortune teller to predict the emotional as well as psychological repercussions in the hearts and minds of the young generation born and raised in Gecekondu society. When they reach their adolescent age, which is the self-realization age, they will see and understand that their way of living is far inferior to that of the city sur- rounding them. These children, unlike their parents, will not consider their shacks adequate and their values no longer will be acceptable ones.13 The crucial problem is that the social behavior of this younger generation neither reflects the older traditional culture nor has it adopted the modern metropolitan culture. llE. T. Gursan, "Gecekondu Cocuk Sagligi," Milliyet Gazetesi, Istanbul, October 1, 1966. 12Tutengil, op. cit., p. 93. l3Celal Uzer, "Gecekondu Problemimiz," Milliyet Gazetesi, Istanbul, August 5, 1964. One may term it an "emerging culture," somewhat between the old and new or a mixture of the two. If the child's back- ground culture fits into the school environment, most prob— ably he can easily adjust to the expectation of school culture. Otherwise, he will fail to cope with the school requirements and he may perceive himself as incompetent in his academic achievement as well as in his socialization. So far, there has been no comprehensive research study of the academic achievement of these Gecekondu children and their socio-psychological behavior within the context of their academic success, and a cross-cultural comparison with other students who are known to be from well-to-do neighbor- hoods in the same city in Turkey. Therefore, exactly what combinations of social, economic, and socio-psychological factors influence their academic performance is unknown. However, it is known that the students who successfully com— plete each year's schooling vary considerably among the middle schools in Turkey; those middle schools in Ankara that were investigated in this study are no exception. Thus, it is believed that the investigation of some selected socio— economic and socio-psychological factors and their influence on students' academic achievement can shed light on the vari- ation in school outcomes (number of students who pass from one grade to the next) among the schools. Currently, investment in education is considered not only for personal satisfaction but also for the prepara- tion of a socially and economically productive individual. 10 Low productivity of the educational system as a whole results in economic wastage and slows down the rate of national develOpment. In general, the schools in Ankara located in the area of Gecekondu dwellings have a lower rate of completion of grades than those located in the affluent neighborhoods. Despite the seriousness of the problem, there has been no systematic comparative study of the academic achievement of Gecekondu children and chil- dren from middle and upper middle class families at the middle school level. Scientific research findings on the issue are badly needed by educational planners and decision makers. It is to this need that the present study is directed. Limitations of the Study The study is based on a sample of sixth grade stu- dents attending middle school in Ankara, Turkey, during the 1971-1972 school year. The sample includes only those schools in Ankara designated as representing Gecekondu schools and well-to-do neighborhood schools. Hence, while implications for the larger student body throughout Turkey may exist, one must understand that this study is focused only upon those schools selected within the geographical limits of Ankara. Therefore, the transfer of generalizations to other geographical regions or to other grade levels within the same region should be made only if the reader is willing to take responsibility for the validity of such extended generalizations. 11 Assumptions Upon Which the Study Is Based The following assumptions are made as the limiting factors for the purposes of this study: 1. A satisfactory survey instrument was devised for the purpose of determining the attitudes of sixth grade stu- dents regarding their school aspiration, self—concept of ability, and their perception of significant others' eval- uations of their academic ability. The instrument was also constructed in such a way that it elicited as much accurate information about student socio-economic background as could be obtained. 2. The sixth grade students responding to the survey instrument (Student Questionnaire) were able to understand the intent of the instrument and its contents, and responded in a manner truly representing their socio-psychological behavior and reflecting their socio-economic background. 3. The sixth grade students responding to the survey instrument were representative of the student subcultures-- namely Gecekondu culture and middle and upper middle class culture--within Ankara. 4. It was further assumed that students' grades in reading, mathematics, social science, natural science, and foreign language given by their teachers were objective indicators of the students' performance in those subjects. 5. The differences, if any, in academic achievement between the two student populations were due to the differ- ences in the students' socio-economic status and 12 socio-psychological factors. Therefore, no other factors, such as teacher quality, teacher expectation, or physical facilities existing in the schools, were assumed to have any impact upon students' academic success. Definition of Terms For clarity of understanding, the following terms are defined either because of their specialized meaning or because of the operational definition which is used in this particular study. Socio-economic Variables (SES)--This term takes into account father's occupation, father's income, father's edu- cation, educational background of the student, and the residence of the student. SES index is the summated scores which each subject gets from the items in the questionnaire. Father's Education——The level of formal schooling which each student's father has achieved. Father's Occupation—~The occupation in which the student's father is currently employed, which he acquired either through formal schooling or on—the-job training; it is the main source of family income. Father's Income—-The monthly wage in Turkish cur- rency earned by the father or the head of the household who was substituted for the father. Residence-~This index is a sum of the scores obtained from the items referring to facilities that exist at the student's home. 13 Educational Background--This indicates whether the student has graduated from the village elementary school, the town elementary school, or the city elementary school. Socioepsychological Variables (SPS)--This index shows the summated scores from the child's school aspira- tion, self-concept of ability, and perceived evaluation by others--namely parents, teacher, and friends. School Aspiration--This indicates how far a child wants or plans to go in his schooling. Self-Concept of Ability--The perception the student has of himself concerning how far he can succeed in a par- ticular performance compared with others. Perceived Evaluation by Others--One's interpretation and internalization of the expectations of significant others concerning his potentialities. Academic Achievement--The grades of each student for a term (four months) in reading (Turkish) and mathematics, and the G.P.A. in five subjects--social science, natural science, foreign language, reading, and mathematics. Father--The legal father or legal guardian with whom the student lives permanently. If the father is dead, the mother is considered to fill this role. Elementary School--A public school which provides five years of education for children between the ages of six and 14. This formal schooling is compulsory for all children. 14 Middle School--A three-year public school which accepts those who have an elementary school graduation diploma. Overview In Chapter I were presented the statement of the problem, the purpose of the study, the significance of the study, delimitations and assumptions, and definition of terms used in the thesis. Chapter II contains a review of the literature con- cerning the relation of students' achievement to their socio-economic background and socio-psychological factors, with some attention given to the debate concerning the effects of environment and heredity on achievement. Included in Chapter III is the design of the study, including a definition of the population, a description of the sample, and a discussion of the data collection proce- dures used in the study. The analysis of the data and a discussion of the research findings are presented in Chapter IV. Included in Chapter V are the summary, findings, and recommendations for further research. CHAPTER II REVIEW OF LITERATURE Introduction This review presents the findings of selected studies focusing on the relationship between students' socio-economic backgrounds and their academic achievement. Because no studies relating Turkish students' socio—economic backgrounds and socio-psychological factors to their achieve- ment were found, this review is limited to the literature and research findings in American publications. The relationship between students' socio-economic backgrounds and their academic achievement has been studied extensively in American literature. Instead of exhaustively reviewing the whole literature pertinent to this study, only a sample of selected studies and literature has been reviewed in this chapter. An Overview of the Debate on the Issue of Environment Versus Heredity Within the Context of Human Learning The capacity of a child to learn has involved a seem- ingly unending debate between those psychologists and educa- tors who stress genetic endowments and those who stress environment as the primary determining factor. Although it is beyond the scope of this study to exhaustively analyze 15 16 the literature of this debate about environment versus heredity, the reader should be aware of some highlights on the issue. Geneticists tend to emphasize that individual variations in learning or variations in the ability to per- form certain tasks is attributable to differences in intel- ligence, which has been assumed to be the inborn capacity of the human being to learn.1 Jensen, a leading contemporary Spokesman of this school of thought, accounts for the total variance in the population in terms of the proportions of the variance attributable to genetic and environmental com- ponents. Jensen believes that social scientists underesti- mate the genetic basis of intelligence.2 The brain mechanisms which are involved in learn- ing are genetically conditioned just as are other structures and functions of the organism. What the organism is capable of learning from the environment and its rate of learning thus have a biological basis. On the basis of his own and his supporters' studies about intelligence versus environment, Jensen recommends the following educational policy: If diversity of mental abilities, as of most other human characteristics, is a basic fact of nature, as the evidence indicates, and if the ideal of universal education is to be successfully pur— sued, it seems a reasonable conclusion that schools and society must provide a range and diversity of educational methods, programs, and goals, and occu- pational opportunities. Diversity rather than uni- formity of approaches and aims would seem to be the lJensen, op. cit., p. 17. 21bid., p. 29. 31bid., p. 45. 17 key to making education rewarding for children of different patterns of ability.4 The other school of thought, to which many other psychologists belong, believes that the performance (intel- lectual, physical, or social) of any individual is not develOped from a genotype, inherited base. Cronbach points out that what the person does with an experience, and what it does to him, depends significantly on his previous exper- ience.5 He further adds that "human development is a cumue lative, active process of utilizing environmental inputs, not an unfolding of genetically given structures."6 Cronbach also disagrees with Jensen on the question of whether intel- 1igence tests really measure inherited factors, defined by Jensen as "9" factors. Cronbach points out that the verbal intelligence test scores of an individual can only reflect the achievement of that individual. Finally, contrary to Jensen, Cronbach states his educational policy as follows: The educator's job is to work on the environment. . . . Heritability of individual differences is not our concern. Even if ranking in ability were to cor- relate perfectly with some measure on pupils' ancestors, the educator ought to be providing the best possible instruction he can for every pupil he faces. On the last point, of courSe, Jensen would no doubt be in agreement. 41bid., p. 117. 5Lee J. Cronbach, “Heredity, Environment and Educa- tional Policy," Harvard Educational Review, XXXIX (Winter, 1969), 338-347. 61bid., p. 338. 7Ibid., p. 345. 18 Brembeck discusses the problem within the context of the school curriculum as follows: Although educators are aware of the importance of social factors in learning, frequently educational programs are based on the assumption that each child has a fixed capacity and that this capacity can be identified and measured. Related to this assumption is the idea that students with low intelligence can not learn at a high level. Often programs are organ- ized to provide the low IQ child with a less difficult curriculum. Such programs constitute a self-fulfilling prophecy in that the students in these special pro- grams will not learn more advanced subjects. If, however, schools were oriented to the theory and findings of research on the issue of how to enlarge the child's learning horizons, then programs would be developed to enhance the abilities of all students to the maximum rather than to limit learning opportunity on the basis of an assumed level of fixed capacity.9 Kerckhoff points out that the ability to perform school tasks is heavily influenced by the child's preschool experience.10 It is therefore a highly controversial matter whether differences in learning ability should be viewed as "given" in the sense that they are inborn, or whether they should be viewed as the result of the child's previous experience, or both. 8Cole S. Brembeck, Social Foundatign of Education (New York: John Wiley and Sons, Inc., 1967), p. 83. 9Ibid. 10Alan C. Kerckhoff, Socialization and Social Class (Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1972), p. 129. 19 For present purposes, it is the hypothesis of this study that, from the perSpective of the school at least, achievement levels vary by differences in environmental factors and that from the beginning, variations in pupil performance at school are significantly related to school influences. Introduction to Related Studies Considerable recent research in the United States has focused on the problem of differences in academic achievement of students and the relation of those differ- ences to their socio-economic or family background and to the behavior they learned within the different subcultures. The purposes and the nature of such studies are varied, but the concern here is only with those studies investigating factors such as parental educational attain- ment, family income, and parental occupation and the effects of these factors on the child's school achievement. Some research findings about differences in students' behavior resulting from their different socio-economic backgrounds and the effects of such behavior on the student's academic achievement will also be reviewed. The evidence obtained from research has indicated that the student's family background and student composition of the schools have played a very important role in stu- dents' academic achievement as well as in the development of behavior. For example, Hollingshead points out that 20 lower-class youngsters have limited their horizons to the class horizon, and in the process they have unconsciously placed themselves in such a position that they will occupy the same levels as their parents.11 Children with differ- ent home backgrounds bring to school differently develOped attitudes and skills. The child's behavior pattern is learned through his interaction with his environment-- through the process of "socialization." Blumer explains the process of interaction among people by "symbolic interac- tion." Symbolic interaction is a social product and it is formed in and through the defining activities of peOple as they interact.12 Blumer gives a brief sketch of Mead's analysis of social interaction: Mead's concern was predominately with symbolic interaction. Symbolic interaction involves interpre- tation, or ascertaining the meaning of the actions or remarks of the other person, and definition, or con— veying indications to another person as to how he is to act. Human association consists of a process of such interpretation and definition. Through this process the participants fit their own acts to the ongoing acts of one another and guide others in doing so. Therefore, it is quite common to see different forms of the socialization process across the subcultures within a society. The effect of interaction of family members on 11A. B. Hollingshead, Elmtown's Youth (New York: John Wiley and Sons, Inc., 1949), pp. 282-287. 12Herbert Blumer, Symbolic Interaction (Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1969), p. 5. 13 Ibid., p. 66. 21 a child's socialization, particularly in early years, is greater than other factors in the social environment in which the child grows up. As Rosen points out, sOcialization ordi- narily begins with a matrix of relationships which can be specified by reference to the roles of individual family members.14 Parents transmit values to their children through constant relationships. Thus, the family is primary in shap- ing the child's personality. Family influence is greatest in the early years, more formative than any other "twig bender."15 In later childhood and youth, a number of forces-- peer groups, the school, mass media, etc.—-compete with and sometimes displace the home as the dominant socializing influence on the child. Review of Related Research Findings Studies relevant to the problem at interest have been extensively carried out, particularly in the United States. But there has been no comprehensive study which deals with this question in Turkey, and no study of socio-economically deprived students in Turkey as related to their academic performance. The research findings have been presented in this study under the heading of each socio-economic (SES) and l4Bernard C. Rosen, "Family Structure and Value Transmission," in Society and Education, ed. by Havighurst, Neugarten and Falk (Boston: Allyn and Bacon, 1967), p. 86. 15Lloyd A. Cook and E. F. Cook, A Sociological Approach to Education (New York: McGraw-Hill, 1960): p. 175. 22 socio—psychological (SPS) factors, rather than by summarizing each study. Therefore, the same research may be referred to more than once under the heading of each factor in this review. Socio-Economic Factors (SES)l6 Although one can enumerate a number of factors which may contribute to the socio-economic status of a child, there seems to be common agreement to use parents' educational attainment (mostly father's), parents' occupation, and family income as the main SES factors which contribute to the child's academic performance at school. From the point of view of preceding explanation about SES factors, we can say that the student's family background differences are prior to school influences, and shape the child before he reaches school. It is useful, then, to examine the relation of student's socio-economic status and his academic achievement before looking at the other factors. Parents' Educational Level and Student Achievement Parental education frequently is chosen as the sole indicator of the social and economic status of a child. McClelland studied some nonintellectual variables and their relationship to college academic achievement in a sample of 16SES and family background of students are used synonymously in this study. 23 233 United States-born freshmen male subjects attending Tri-State College, Angola, Indiana.17 In this study, cumu- lative GPA was used as the predicted variable, and the data were analyzed by applying multiple correlation. The find- ings indicated that the attained educational level of parents and student's performance at college were correlated sig- nificantly. Mayeske has develOped some indices of students' family background by grouping variables used in the "Equality of Educational Opportunity" survey.18 He found that a stu- dent with a high score on his SES index has parents who come from the higher educational strata, and his father is typi- cally engaged in a professional, managerial, or skilled job. Hood investigated the educational and personality factors associated with parental education.19 The study was designed to investigate the general nature of the relation- ship between parental educational level and certain educa- tional characteristics of children. His findings showed that parental educational status is more related to plans for l7McClelland, op. cit., p. 44. 18G. W. Mayeske, "On the Explanation of Racial Ethnic Group Differences in Achievement Test Scores" (Washington, D.C.: U.S. Office of Education, n.d.), p. 23. (Mimeographed.) 19A. B. Hood, "Educational and Personality Factors Associated With Unusual Patterns of Parental Education," Journal of Educational Research, LXI (1968), 32. ‘J [I]! . 24 children to attend college than is family economic status among high school students. Lowe has investigated some selected socio-economic factors and their relation to seventh grade students' read- ing performance in Virginia.20 He concluded that reading scores were higher for students whose parents had obtained high levels of education than others whose parents had not. The correlation was .30 with mother's education and .33 with father's education. The report of "Equality of Educational Opportunity" is one of the most comprehensive studies about the relation- ship of socio-economic factors to students' academic achievement.21 Analysis of data, obtained through a stu- dent questionnaire, indicated that in the sixth grade the parents' educational level had made a higher contribution to student's achievement for white pupils than for any other groups. But, in later years, parent's education comes to have the highest relation to achievement for nearly all groups of students. 20Walter E. Lowe, "A Study of Relationship Between the Socioeconomic Status and Reading Performance of Negro Students Enrolled in the Public Schools of Caroline County Virginia," The George Washington Universipy Bulletin, Abstracts of Doctoral Dissertations, LXIX, 1 (September, 1968), 36. 21J. 8. Coleman, quality of Educational Opportunity (Washington, D.C.: U.S. Government Printing Office, 1966), pp. 298-302. 25 Parents' Occupation and Student's Academic Achievement It is difficult to factor out the unique contribu- tion of parents' occupational status independent from other SES factors related to the child's academic achievement, since parents' occupational level (mostly father's occupa- tion) and family income and parents' educational level are highly correlated with each other. For example, with a higher educational attainment a man is more likely to have a professional or managerial job. Most children are trained by home and neighborhood to occupy the social position of their parents. The schools in the lower—class environment offer some conflict to home and neighborhood training, but the consequence is usually a losing battle on the school side. On the other hand, the school program supports and supplements the home and neighborhood training of middle- class children.22 Knieff and Stroud studied the intercorrelation among various intelligence, achievement, and social class scores of 344 fourth grade pupils in a Midwestern city in the United States.23 They used father's occupation as a social class index, and a significant relationship was found 22W. L. Warner, Robert J. Havighurst, and M. B. Loeb, Whoghall Be Educated (New York: Harper and Brothers, 1944) I po 56. 23L. M. Knieff and James B. Stroud, "Intercorrela- tions Among Various Intelligence, Achievement and Social Class Scores," Journal of Educational Psychology, L, 3 (1959), 117-120. 26 between father's occupation and pupil's academic achieve— ment. A study conducted by Wilson in Richmond, California, assessed the relationship between a student's social class and his school achievement.24 In this study, parents' occupational level was used as the social class index of the student. The research findings indicated that parents' occupation was the single factor most related to the aca- demic achievement of children. Mayeske found that the lower the level of fathers' occupations, the lesser the mean achievement scores of their children.25 High occupational level usually goes with high income and vice versa. Since parents' income and parents' occupational level are highly correlated with each other, the variation in pupils' achievement accounted for by par- ents' income independent from other SES factors mentioned before is not great. However, it is worthwhile to look at some studies which deal with the relationship between pupils' achievement and family income. ‘ Family Income and Students' Academic Achievement There are some general characteristics associated with the relative poverty or affluence of the family which 24Alan B. Wilson, The Consequences of Segregation: Academic Achievement in a Northern Community (Berkeley: University of California Press, 1969). 25G. W. Mayeske, A Study ofygur Nation's Schools (Washington, D.C.: U.S. Government Printing Office, 1969), p. 29. 27 are related to success or failure of students in school. Coster has investigated some characteristics of 878 high school pupils from different income groups in nine Indiana high schools.26 The study was concerned primarily with pupils' participation in social activities and secondarily with grades according to their family income. In this study, students were divided into three income groups for the anal- ysis of data. From the research findings, it was concluded that pupils from high-income families were more likely than middle- and low-income pupils to participate in school activities and to get higher grades on academic performance. Davres did a comparative study of the performance of pupils from low, high, and economically diversified socio- economic areas in Kansas City, Kansas}.7 The purpose of the investigation was to find out how well students representa- tive of low-income families achieve in comparison to students representative of higher income families when achievement is measured from test items. The study was limited to the ninth grade students of Kansas City. Findings indicated that chil— dren from low-income families did not achieve as well as the other pupils representative of high-income families on a standardized achievement test of social studies. The scores 26J. K. Coster, "Some Characteristics of High School Pupils From Three Income Groups," Journal of Educational Psychology, L, 2 (1959), 55-62. 27W. L. Davres, "A Comparative Study of the Perform— ance of Pupils From Low, High and Economically Diversified Socio-Economic Areas on Test Items From a Social Studies Achievement Battery" (unpublished Ph.D. dissertation, Univer- sity of Kansas, 1967). 28 of children representative of low-income areas were signif- icantly lower than scores of children representative of a high-income area. General SES Factors and Achievement There are a number of factors which may constitute the socio-economic background of a child within a particu- lar culture. Each factor can influence the child's behavior and indirectly those factors affect his academic achievement. The Coleman report examined cross-culturally the influence of students' socio-economic backgrounds on their achievement. The variance in achievement accounted for by students' background was between 30 and 50 per cent of the total variance for all groups.28 Mayeske did a communality analysis with the data used for "Equality of Educational Opportunity" to find out how much variation in students' achievement can be accounted 29 The statistical analysis for by combined SES factors. revealed that 51 per cent of the total variation in academic achievement was due to the differences of students' socio- economic background. Jencks and his associates reassess the findings of some of the most recent research about the influence of economic background on students' achievement. He concludes 28Coleman, op. cit., p. 299. 29Mayeske, "On the Explanation of Racial Ethnic Group Differences in Achievement Test Scores," op. cit., p. 17. 29 that: On almost any reasonable set of assumptions, family background explains nearly half the variation in edu- cational attainment. A family's economic status is, of course, a major determinant of its overall impact on its children. But noneconomic factors also account for a significant fraction of 3 family's overall effect on its children's attainment. 0 Thus, the overall review of literature indicates that a student's family background has a substantial effect on his academic performance at school, and, in fact, that school factors themselves do not have as much effect as do the student's socio-economic background factors. Socio-Psyghological Factors (SPS) Differences in human behavior, including school per- formance, are much more related to differences in the social environment than to differences in the physical environment. In the early shaping of a child's life, home has a signifi- cant effect on personality formation. The family's impact on the child has its greatest effect in the earliest years.31 In the second phase of a child's life, school becomes an agency for the develOpment of self, particularly of his perception about himself and evaluation of others with respect to his education and learning. In general, the impact of the family is greatest and most completely unchallenged in the preschool years, 30Christopher Jencks, Inequality: A Reassessment of the Effect of Family and Schooling in America (Boston: Basic Books, 1972), p. 143. 31Coleman, op. cit., p. 300. 30 lessening as the child gets older. The social develOpment of the child at the age of five or six is such that the school tends to be a more significant influence than other environmental factors, next to the home. Differences in motivation, values, social environments, and characteris- tics patterns of children have a significant relation to school achievement as well as to the desire for further edu- cation. Therefore, the following social-psychological fac- tors in relation to a child's academic achievement have been reviewed. School Aspiration and Achievement "Aspiration for education" indicates how far a child wants or plans to go on with his schooling. Brembeck points out that aspirations for further education are nurtured within a social context.32 There are a variety of social experiences which stimulate educational aspirations. These stimulating factors are very closely associated with the socio-economic status of a child. What children reflect in classroom behavior are the norms which they have learned in their culture. Children bring to school with them a predisposition to behave as they do in out-of—school groups.33 32Cole S. Brembeck, "Raising Educational Aspirations and School Learning," in Social Foundations of Education, ed. by Brembeck and associate (New York: John Wiley and Sons, 1969), pp. 263-288. 33 Brembeck, Social Foundation of Education, op. cit., p. 87. 31 Many studies in the United States give considerable weight to the influence of social class in the determination of educational aspirations. In his study "Academic Achieve- ment in a Northern Community," Wilson found a strong rela— tionship between students' school aspirations and their social class status.34 In the study "Equality of Educational Opportunity," it was found that the pupil's attitude toward education has a stronger relationship to achievement than all the other school factors combined.35 Students' school aspirations toward further education are primarily a result of home and school influences. It is obvious that if the home and school environment do not motivate the pupil toward learning, it can hardly be expected that a child will achieve at school. Coleman found that a smaller proportion of black than white students reported wanting to go further than high school. Also, fewer blacks have definite plans for college and less consistency regarding school aspirations than whites.36 In a study of social aspects of aspirations in the public schools of Berkeley, California, Wilson found that children of higher social status showed higher aspirations and achieved more than did children of lower status. 34Wilson, op. cit. 35 Coleman, op. cit., p. 23. 36Ibid., p. 279. 32 School aspiration, in fact, was found to be closely related to student's family background. However, the "significant others" (see below) play the major role in the process of inspiring him to value and attain education.37 Significant Others Significant others refers to the influence of fam- ily members, playmates of the child, classmates, friends, and teachers. A child's original images of himself are formed in the family circle. As he grows up his friends, his class— mates, and his teacher join this circle. The individual forms and aligns his own actions based upon his interpreta- tion of the expectations, acts, and opinions of others. The foregoing are the essential features, as Blumer sees them, in Mead's analysis of the bases of symbolic inter- action.38 Human beings respond to one another on the basis of the intentions or meanings of gestures. As Meltzer interpreted it from the point of view of "Mead's Social Psy- chology," this gesture becomes a symbol to be interpreted in the imagination of the participants.39 Through symbolic interaction and communication with others significant to him, the child develops his self. A 37Alan B. Wilson, "The Effect of Residential Segrega- tion Upon Educational Achievement and Aspiration (unpublished Ph.D. dissertation, University of California, Berkeley, 1960). 38Blumer, Op. cit., p. 82. 39Bernard N. Meltzer, The Social Psychology of George Herbert Mead (Kalamazoo, Michigan: Division of Field Ser- VICes, Western Michigan University, 1959), pp. 11-15. 33 child's self-concept of ability, which has been found as a significant factor correlated with achievement, is acquired during his interaction with significant others who hold expectations of the child as a learner. The impact of others' expectations and evaluations on the student's behavior have been investigated extensively in the literature. Brookover and his associates investigated some selected socio-psychological behavior of secondary school students in Michigan and the relationship of that behavior to school achievement.40 The findings indicate that parental evaluations of their child's academic ability were more related to his self-conceptions of academic ability than were friends' evaluations of his academic ability in grades seven, eight, nine, and ten. From grades seven through twelve, the impact of parental evaluations on self-concept of ability was greater than that of teacher evaluations. Brookover points out that if there is a general and homogeneous set of high expectations held by all significant others (parents, friends, and teachers) for a child, then a relatively high level of academic achievement could be expected.4 Sidawi investigated socio-psychological variables and their relations to the academic achievement of Lebanese 40W. B. Brookover, Edsel Erickson, and Lee M. Joiner, Self-Concept of Ability and School Achievement, III. U.S. Office of Education Cooperative Research Project No. 2831 (East Lansing: Educational Publication Services, Michigan State University, 1967), pp. 107-109. 41 Brookover and Erickson, op. cit., p. 93. 34 children during junior high school.42 He found that per- ceived evaluation of parents, friends, and teachers is strongly related to a child's self-concept of academic abil- ity. School aspirations, child's perceptions about the expectation of significant others, and self-concept of abil- ity seemed to be highly intercorrelated with each other. Therefore, it is believed that reviewing the literature on self-concept of ability may shed light on why one student is a better achiever than another. Self-Concept of Abilitygand Achievement The increasing recognition of the importance of "how a child views himself" has been followed by a variety of theoretical descriptions of the nature and influence of the child's self—concept on other aspects of his development. As McCandless points out, the self-concept may be thought of as a set of expectancies, plus evaluations of the areas of behavior with reference to which these expectancies are held.43 Perkins describes self—concept as those perceptions, beliefs, feelings, attitudes, and values which the individual views as describing himself.44 42Ahmad Sidawi, "Self-Concept of Ability and School Achievement in Lebanon" (unpublished Ph.D. dissertation, Michigan State University, 970). 43B. R. McCandless, Children and Adolescents (New York: Holt, Rinehart and Winston, 1961). 44H. V. Perkins, "Factors Influencing Change in Chil- gggn's Self-Concepts," Child Develgpment, XXIX (1958), 221- 35 The conception that a person forms of himself usually has a social reference: generally, it takes the form of the "self" system a person acquires in the course of socializa- tion and depends largely on the kinds of personalities with which the person is associated. Self-concept is develOped in part through social interaction. Thus individuals who have different experiences in interacting socially will have different self-concepts. Klausner found in his study of "Social Class and Self-Concept" that there are modal differences in self- concepts between members of different socio—economic group- ings, and that members of the same socio-economic grouping tend to have a relatively homogeneous self-concept.45 Self-concept of ability is just one of many concepts of self. In this study, the concern is focused on areas of behavior relating to school achievement. Thus, for our pur- poses, self—concept of ability refers to what one expects to achieve in academic tasks as compared with others engaged in the same task. There is ample research evidence that the student's academic achievement is closely related to his self-concept of ability.46 Brookover and his associates have carried out a ser— ies of researches on the problem of self-concept of ability and school achievement. Brookover's 1967 report has the 4SS. Z. Klausner, "Social Class and Self-Concept," Journal of Social Psychology, XXXVIII (1953), 201—205. 46 Brookover, Erickson, and Joiner, op. cit. 36 following findings: 1. The correlation between socio-economic status and self-concept of ability was .26 for eighth grade and .23 for ninth grade. The correlation between self—concept of ability and G.P.A. was .55 and .56 for grades eight and nine, reSpectively. Improvement in students' socio-economic status over a five-year period yielded slight increases in the self-concept of ability and G.P.A. at each grade level. Using partial correlation, controlling for variation in self-concept of ability, the relationship between socio-economic status and G.P.A. was reduced to near zero at each grade level. The correlation between self-concept of ability and G.P.A. ranged from .48 to .63 over the five years (between 1962 and 1967). A high correlation between perceived evaluations and self-concepts was found. It was concluded that perceived evaluations (parents, friends, and teach- ers) are a necessary and sufficient condition for self-concept of ability, but self-concept of ability is only a necessary but not sufficient condition for achievement. 47Ibid. 37 Summary In this chapter, pertinent research literature on student's socio-economic background (family background) and his socio-psychological behavior with relation to his aca- demic achievement was reviewed. The amount of literature dealing with a student's background and its effects on his achievement is considerable. Most of the writings focus on the central theme that the student's socio-economic background has an indirect effect on his academic achievement; and the key to the main factors which have impact upon academic achievement is the student's developed socio-psychological concepts about him- self as well as his attitudes toward learning. Socio-psychological factors seem to be the most influential on the student's academic achievement. Economic factors seem to be less important than the attitudinal ones. However, it is difficult to specify precisely the ways in which the student's physical and social environment affect his attitudes and his academic performance. There are com- plicated relationships among socio-economic and socio- psychological factors and academic achievement. From the reviewed literature, one thing is very clear--that children who grow up in socio—economically deprived areas frequently develop a negative attitude toward schooling, and conse— quently they become lower achievers in an academic task than others who come from high SES families with positive attitudes toward education. The attitudes of children toward 38 their school work are deeply affected by the degree of encouragement from the significant others (parents, teachers, friends) and by their self—conception of ability. The present study attempts to determine if these relationships hold for the Turkish culture. The methodology used in this study is presented in Chapter III. CHAPTER III RESEARCH METHODOLOGY The primary purpose of this study was to determine the influence of selected socio-economic and socio- psychological factors on academic achievement of sixth grade children in Turkey during the 1971—1972 school year. Secon- darily, the purpose also was to compare two types of student populations with respect to differences, if any, in rela- tionships of those above-mentioned nonintellectual variables with achievement. This chapter describes the pOpulation of interest, sampling procedure, instrumentation, and the tech- niques for analysis of data. Definition of the POpulation The pOpulation in this study consisted of two strat- ified student populations attending sixth grade in Ankara, Turkey, in the 1971-1972 school year. One of the strata includes four schools (Akdere, Safaktepe, Gulveren, and Aktepe) known to be attended by students from Gecekondul (low SES) areas. The second stratum consists of three schools (Namikkemal, Mimarkemal, and Bahcelievler) located in well-to-do neighborhoods (high SES) where most of the 1Gecekondu refers to a typical housing which is poorly constructed and poorly furnished, and usually it consists of one or two rooms. 39 40 students come from middle- and upper—class families. The sixth grade pOpulation in these seven public middle schools ranges in size from 500 to 1,250. Almost 34 per cent of this student population is between eleven and thirteen years old. All seven schools are coeducational, and all students are exposed to the same curriculum prepared by the Ministry of Education. There is no grouping within the classes of these seven schools with respect to any kind of individual ability. The Sample In this study, the sample is a stratified cluster sample in which the class and not the student is the primary sampling unit. By the application of the cluster sampling technique, two classes from each of six schools and four classes from one school were selected randomly and prOpor- tionally with respect to sixth grade populations in each school. The sample consists of 378 students, 219 male and 159 female, from non-Gecekondu schools and 364 students, 245 male and 119 female, from Gecekondu schools. Thus, the total sample size is 742 students, of which 62.5 per cent are male and 37.5 per cent are female. The total number of sixth grade students enrolled in these seven middle schools was approximately 5,000 students in the 1971-1972 school year. 41 Sources of the Data Data for the study were obtained from two sources: the school record for each student, and a "Student Ques- tionnaire." Grades of each student in reading and mathe- matics and a combined G.P.A. on five subjects——reading, mathematics, social science, natural science, and foreign language-—were obtained from students' files at school. The information about the student's socio-economic background and his perceptions about himself with respect to his academic future and his educational aspirations were obtained by means of a student questionnaire which was supplemented by a parent questionnaire. Description of the Instrument A measurement device, entitled "Oggenci Anketi"3 (Okul—Cevre Arastirmasi) translated "Student Questionnaire" (School Social Environment Study) was prepared by the researcher. Some of the items in the questionnaire related to the student's self-concept of academic ability, his school aspirations, and the perceived evaluations of the student's ability by parents, friends, and teachers. These items were translated into Turkish and adapted with minor changes from 2"Parents Questionnaire" is a supplementary device to the Student's Questionnaire. It consists of six open- ended items which were filled in by parents at home and then students brought it back to school and used the infor- mation in it when responding to the same types of items in the Student Questionnaire. 3See Appendix A. 42 the items of the "Student Questionnaire" which was used for the School Social Environment Study by BrookOver at Michigan State University.4 Other items concerning a student's socio-economic background, including his educational background, father's education, father's occupation, father's income, and his residence conditions, were prepared by the researcher. These items were discussed with Turkish students attending Michigan State University in the 1971-1972 academic year. The sug- gestions of these students were incorporated into the first revised draft of the instrument. It was then reviewed by Michigan State University professors in the areas of educa- tional research. Based on their comments and suggestions, the instrument went through further revision. The completed survey instrument (Student Question- naire) was pilot tested by administration to seven fifth and sixth grade Turkish students in East Lansing, Michigan, in January, 1972, to determine the clarity of meaning and style of wording. As a result of the pilot testing, further minor revisions were made to improve the clarity and simplicity of several questionnaire items. The final draft of the instrument was sent to PAKD (Planlama-—Arastirma ve Koordinasyon Dairesi) in Turkey. After minor changes were made on some items by the eXperts of PAKD, the Student Questionnaire was printed. The printed 4Student Questionnaire (revised draft), School Social Environment Study, sponsored by Michigan Department of Educa— tion and Michigan State University. 43 instrument consisted of 52 items, of which the first 17 items were designed for getting information about the stu- dent's identification and his socio-economic background. The next nine items pertained to the student's school a5pirations and his self-concept of ability. The 18 suc— ceeding items were about the student's perceived evaluation of his ability by parents, teacher, and friend. The final six items dealt with the academic climate among students and teachers, and some miscellaneous matters. The supplementary questionnaire (Parent Question- naire), which consists of six open-ended questions concerning basically parent's education, occupation, income, and the type of residence, was also constructed by the author of this study.5 Since the questionnaires were to be administered by someone in Turkey in the author's absence, detailed instruc- tions were prepared for administering the questionnaires and for getting the academic achievement grades of those subject areas specified for each student from his file at the school. All of the data collection devices and instruc- tions for administration of those devices were mailed to PAKD in March, 1972. Collection of the Data Both questionnaires described above were administered by members of the PAKD staff in May, 1972, in a two-step 5See Appendix A. 44 administration. In the first step, each student whose class was included in the sample was given a "Parent Questionnaire," which is self—explanatory. The items in the parent question- naire were to be answered by their parents and the question- naire was to be brought back to school by the student on the day that the Student Questionnaires were planned to be admin— istered. The student questionnaires were group—administered in their classrooms by PAKD staff members with the coopera— tion of school directors and teachers. The students were told that they could use the information provided by the Parent Questionnaire while they were answering the similar types of items in the Student Questionnaire. The completed questionnaires and the grade reports of the students included in the sample were mailed to Michigan State University in May, 1972, and data were received by the author without loss. A summary distribution of students, by schools par- ticipating in this study, in presented in Table 1. After a "Coding Sheet"6 was prepared by the author with the advice of professors of his academic committee, each item in the student questionnaire was scored and coded by the researcher in such a way that the magnitude of the student's response to each questionnaire item would be con- sistent with the prospective statistical treatment of the 6See Appendix B. 45 data. Because of misprinting, the items numbered 31 and 46 on the questionnaire were dropped. Frequency of nonresponse to the items with reSpect to total response was negligible. As a measure of accuracy, each child's response with the corresponding response of his parents was checked. After coding was completed the double checked, the information was transferred onto IBM cards, one card for each student, containing his school identification. Table l.--Distribution of subjects by school in the survey. —..... Number of Number of SES Category Name of School Cluster Students of School Namik Kemal 2 90 non-Gecekondu Mimar Kemal 4 215 non-Gecekondu Bahcelievler 2 73 non-Gecekondu Subtotal 8 378 Akdere 2 92 Gecekondu Safaktepe 2 95 Gecekondu Gulveren 2 76 Gecekondu Aktepe 2 101 Gecekondu Subtotal 8 364 Grand Total 16 742 Statistical Treatment of the Data The study was essentially concerned with obtaining data that would be used in answering the following research questions: A. Questions for combined populations 1. What is the magnitude of the relationship, if any, between a student's academic achievement and his socio-economic status? 46 2. What is the magnitude of the relationship, if any, between a student's academic achievement and socio—psychological variables? 3. Is SES or SPS more significant in establishing these relationships? B. Questions for comparing the two sub-populations 4. To what extent, if any, do SES and SPS variables differ, between the two student populations, in predicting students' achievement: a. in reading b. in mathematics c. in G.P.A. of five subjects--reading, mathe- matics, social science, natural science, and foreign language? The survey instrument was designed so that the responses of the sixth grade students on different parts of the instrument could be compiled and used as a basis for answering the research questions specified above. The statistical treatment of the data may be summar- ized as follows: 1. After the information was transferred onto IBM punch cards, frequency counts, percentages, arithmetic means and standard deviations, where appropriate, were computed for each item on the questionnaire by using the CDC 6500 CISSR Percount Program available at Michigan State University. 7Larry Thiel and Linda Patrick, Percount, Technical Report No.,18 (East Lansing: Michigan State University Com- pu er Institute for Soc1al SCience Research(CISSR), 1 68). 47 2. Groups of items from the Student Questionnaire were combined to develOp indices such as self-concept of ability and student's perceived evaluation of his academic ability by significant others (parents, teacher, and friend). Before these Likert-type indices8 (summated scores procedure) were analyzed, Hoyt's9 reliability was computed for the items of each index to see whether those items were internally con— sistent with each other by using CDC 6500,the FORTAP,program available at Michigan State University.10 3. In step three of the analysis of data, the inter- correlation of 63 items (included five indexes) was computed by using CDC 6500,the BASTAT,program available at Michigan State University.11 4. In the fourth step, factor analysis was applied to the data in order to identify groups of variables that not 8Rensis Likert, "The Method of Constructing an Atti- tude Scale," in Readings in Attitude Theory and Measuremgnt, ed. by M. Fishbein (New York: John Wiley and Sons, 1967), pp. 90-95; F. N. Kerlinger, Foundations of Behavioral Research (New York: Holt, Rinehart and Winston, Inc., 1964): pp. 484-488. 9C. J. Hoyt, "Test Reliability Estimated by Analysis of Variance,“ in Principles of Educational and Psychological Measurement, ed. by W. A. Mehrens and R. L.fEbe1 (Chicago: Rand McNally Co., 1967), pp. 108-115. (See Appendix A for further information on Hoyt's reliability.) 10F. B. Baker and T. J. Martin, FORTAP, A Fortran Test Analysis Package, Occasional Paper No. 10 (East Lansing: Office of Research Consultation, College of Education, Michigan State University, 1970). 11Tom Carroll, Marylyn Donaldson, and Leighton Price, BASTAT in STAT Prqgram CDC 6500 (East Lansing: Michigan State University, Agricultural Expefiment Station, 1970). 48 only correlated substantially with one another but were also psychologically or sociologically meaningful. By using the CDC 6500 CISSR FACTORA program,12 similar kinds of corre— lated variables were grouped into SES and SP8 indices. 5. In order to explain the differences in achievement of specified academic areas due to the differences of stu- dents' socio-economic background and to socio-psychological factors or combination effects of both on achievement, the techniques of regression and stepwise regression analysis were employed by using the CDC 6500 LS program and the CDC 3600 LSADD program in STAT program available at Michigan State University.13 The purpose in using regression and stepwise regres- sion analysis was to provide more refined measures for exploration of differences in academic achievement by util- izing more than one variable at the same time, and to iden- tify the factors or combination of the factors which best explained the variance. Summary The procedures, instrumentation, and methodology employed in gathering and analyzing data for the study were described in this chapter. There were two primary sources 12Leighton A. Price and Gary R. Ingvaldson, FACTORA, Technical Report for CDC 6500, Principal Components Factor Analysis (With Orthagonal Rotations) (East Lansing: Michi- gan State University, 1970). 13Carroll, Donaldson, and Price, op. cit. 49 of the data for the study: school records for student's academic achievement and the Student Questionnaire admin- istered by the staff of PAKD in Turkey. The survey instruments (Student Questionnaire and Parent Questionnaire as a supplementary device to the first one) were constructed by the researcher and pilot tested. Parametric statistical techniques were used in analyz- ing the data obtained for the study. These procedures included the use of descriptive summaries of item responses, in terms of frequency counts and percentages, and selected further analysis of data through the use of such statistical techniques as correlation matrix, factor analysis, and multiple-regression analysis. The analyses were carried out on the CDC 6500 and the CDC 3600 computers at Michigan State University. The results of the various data analysis tech— niques are presented in Chapter IV. CHAPTER IV ANALYSIS OF THE DATA The purpose of this study was to determine the mag- nitude of influence of some selected socio—economic and socio-psychological factors on students' academic achievement at the sixth grade in Turkey. In order to do so, specific research questions were posed and relevant data were sought. Chapter IV is divided into the following six main sections: 1. Characteristics of the respondents to the survey 2. Developing indices on selected item and reliability analysis 3. Correlation analysis 4. Factor analysis for developing SES and SP5 indices 5. Multiple regression analysis 6. Stepwise regression analysis Characteristics of Respondents The data presented in this section were obtained from the analysis of sixth grade students' responses to items included in the "Student Questionnaire," the survey instrument used in the study.1 1See Appendix A. 50 51 The 742 sixth grade students responding to the sur- vey were attending public middle schools within the metropol- itan area of Ankara, Turkey, during the school year 1971-1972. The distribution of respondents, by sex and by school cate— gory, is presented in Table 2. Table 2.--Distribution of subjects by sex and by school category. Non-Gecekondu Gecekondu Number of Number of Sex Subjects Per Cent Subjects Per Cent Male 219 58 245 67 Female 159 42 119 33 Total 378 100 364 100 The ratios in the table indicate that the difference of male and female student pOpulation in Gecekondu schools is greater than in non-Gecekondu schools. Female students represent one-third of the total Gecekondu sixth grade pOp— ulation, while the sex ratio in the non—Gecekondu population is almost 50 per cent. The sixth graders responding to this survey repre- sented a considerable range in ages. As indicated in Table 3, one student in each sub-population was less than ten years of age and 40 per cent of non-Gecekondu students and 61 per cent of Gecekondu students were over twelve years of age. Eleven and twelve years of age are known to be nor- mal age for sixth grade in Turkey. Thus, in Gecekondu schools 52 there are more overaged sixth graders than in non-Gecekondu schools. Table 3.--Distribution of subjects by age and by school category. Non-Gecekondu Gecekondu Number of Per Cum. Number of Per Cum. Age Subjects Cent Per Cent Subjects Cent Per Cent 14+ 41 11 11 99 27 27 13 108 29 40 123 34 61 12 195 51 91 131 36 91 11 33 9 100 10 3 100 10 l .. .. l .. .. Total 378 100 364 100 The differences in the age category between the two student populations appear to be due to the different number of students who repeat sixth grade. Table 4 is presented for the better explanation of whether students are attending sixth grade as a first year or second year. Table 4.--Distribution of subject as first year attender or repeater in sixth grade, by school category. Non-Gecekondu Gecekondu Per Cent Per Cent First year attender 85.45 70.05 Repeater 14.55 29.95 Total Subjects N=378 N=364 53 As shown in Table 4, there are more repeaters in Gecekondu schools than in non—Gecekondu schools. As was stated in Chapter I, one of the objectives of this study was to compare the academic achievement of two types of student populations, defined as low socio—economic and high socio—economic groups. By giving the description of Gecekondu dwellings and well—to-do neighborhoods in Ankara, Turkey, it was assumed that students attending the schools in Gecekondu areas represent low socio-economic population and students attending the schools in well-to-do neighborhoods represent high socio-economic populations in this study. Although stratification of student population was based on two types of residential areas, the information about the student's socio-economic background was sought by means of items in the "Student Questionnaire" which relate his father's occupation, father's income, father's education, residence, and his school background. Tables 5, 6, and 7 give the student distribution by their father's occupation, income and education, respectively. As indicated in Table 5, the two student populations differ in their fathers' occupations. The students in the non-Gecekondu population have fathers in more professional and semi-professional occupations than the Gecekondu stu— dent pOpulation. 54 Table 5.-—Distribution of subjects by their father's occupation and school category. Non-Gecekondu Gecekondu Per Cent Per Cent Nonprofessional and low prestige occupation 45 82 Semi-professional and moderate prestige occupation 34 17 Professional and high prestige occupation 21 1 Total Subjects N=378 N=364 The distribution of students by their father's income, as presented in Table 6, indicates that 21 per cent of non- Gecekondu students have fathers who earn more than 2500 T.L. per month, while only 2 per cent of Gecekondu students have fathers whose monthly wage is at this level. Table 6.--Distribution of subjects by their father's income and school category. Non-Gecekondu Gecekondu Income Interval (Turkish Currency) Per Cent Per Cent 0 - 1000 T.L. 45 65 1001 - 2500 T.L. 34 33 2501 and up 21 2 Total Subjects N=378 N=364 Table 7 gives the percentage distribution of students by their fathers' educational level. This table reveals that 55 there are considerable differences between the two popula- tions of students with respect to their fathers' educational level. Cumulative percentages show that in the Gecekondu population only 15 per cent of students' fathers have an education above elementary school and 85 per cent have either an elementary school certificate, or some elementary educa- tion, or none. In the non~Gecekondu population the situation is different, as 56 per cent of students have fathers who have education above elementary school level and 38 per cent of the students' fathers either graduated from high school or have some higher education, while only 5 per cent of Gecekondu students' fathers have achieved this level of edu— cation. If we look at the figures representing the level of some secondary school education--first and second cycle-- 31 per cent of non-Gecekondu students' fathers have had some secondary education, but only 14 per cent of the Gecekondu population has achieved this level of education. The signif— icant difference between the populations with respect to their fathers' attained educational level appears at the university or higher educational level as well. Thus 22 per cent of the non-Gecekondu pOpulation have fathers who grad- uated from a higher educational institute, while only one- half of one per cent of the Gecekondu population attained that level of education. 56 Table 7.--Distribution of subjects by their father's highest educational level and by school category. Non-Gecekondu Gecekondu Highest Educational Cumul. Cumul. Level of Father Per Cent Per Cent Per Cent Per Cent None 9 9 15 15 Some elementary education 6 15 20 35 Graduated from elementary school 29 44 50 85 Some first cycle of secondary educationa 7 51 5 90 Graduated from first cycle of secondary education 7 58 4 94 Some second cycle of secondary educationb 4 62 l 95 Graduated from second cycle of secondary education 13 75 4 99 Some higher education 3 78 .5 99.5 Graduated from a university or a school of higher education 22 100 .5 100 Total Subjects N=378 N=364 aFirst cycle of secondary education indicates three years of education above elementary education. bSecond cycle of secondary education indicates three years of education after first cycle. Table 8 reveals the information about home facilities reported by students in this study. 57 Table 8.--Distribution of subjects by existing home facilities and school category. Non-Gecekondu Gecekondu Home Facility Per Cent Per Cent Telephone 21 2 Central heating 46 2 Gas 67 7 Electricity 100 95 Running water 98 88 Total Subjects N=378 N=364 According to data obtained by the Student Question- naire, the main difference, with respect to home facilities, between the two populations appears on telephone, central heating, and gas. As for electricity and running water at home, the two populations do not differ appreciably. For comparing the two sub-populations with respect to students' given responses to socio-psychological variables, Tables 9 and 10 are presented. Table 9 gives the percentage distribution of students by their school aspiration by school category. This table reveals that the two populations differ according to the level of education that students want to finish, particularly at university level. More stu— dents in non-Gecekondu desire to finish university than do Gecekondu students. Table 10 describes the two populations with respect to SP8 variables—~school aspiration, self—concept of ability, 58 Table 9.--Distribution of students by their school aspira- tion and by school category. , Non-Gecekondu Gecekondu Highest Level of Education Aspired to Per Cent Per Cent Middle school 6 14 Vocational high school 4 14 High school (college prep.) 10 14 University or school of higher education 80 58 Total subjects N=378 N=364 perceived evaluation by others (parents, teacher, and friend)--in terms of means and standard deviations. As indicated in Table 10, non-Gecekondu students have a higher mean than Gecekondu students on school aSpiration, but Gecekondu students show a higher variation than non-Gecekondu students. For self-concept of ability the mean difference between the two populations is small and the variations within each population are almost the same. Except with respect to school aspiration, the data indicate, therefore, that students in low SES group con- sider themselves as able to do the same as those in high SES groups. The mean of perceived evaluations by parents differs across the two student populations but again the difference is not significant. The mean of non-Gecekondu students on this variable is higher than Gecekondu. The differences between means for perceived evaluation by others (teacher and friend) were found to be negligible across the two pOpulations. 59 Table 10.--Means and standard deviations of socio- psychological (SPS) variables by school category. T 6“. o Non-Gecekondu Gecekondu SPS Variables Mean S.D. Mean S.D. School aspiration 3.64 .82 3.17 1.11 Self-concept of ability 21.98 3.18 22.20 3.44 Perceived evaluation by parents 15.18 2.75 14.89 2.77 Perceived evaluation by teacher 16.17 2.96 16.22 2.87 Perceived evaluation by friend 12.56 2.45 12.48 2.51 Total Subjects N=378 N=364 aMeans and standard deviations of self-concept of ability, perceived evaluations by others (parents, teacher, and friend) are based on student's summated scores obtained from more than one item for each SPS variable. Summation procedures are explained elsewhere in the text. Summary of the Section The preceding section of this chapter shows the char— acteristics of two sub-population groups--Gecekondu and non- Gecekondu students-~by means of descriptive analyses of data obtained from the Student Questionnaire in this study. The over-all picture of the two student pOpulations shows that there are more female students in non-Gecekondu schools than Gecekondu schools. Second, in Gecekondu schools there are more over-aged--l3 or above—-students than in non-Gecekondu population. This appears to be due to the high percentage of grade repeaters in those schools. 60 In terms of basic socio—economic variables--father's occupation, father's income, and father's attained educa- tional level--the two student populations differ consid- erably from each other. Thus the non-Gecekondu student pOpulation has fathers who are engaged in more professional jobs, get higher incomes, and have attained a higher educa- tional level than the Gecekondu student population. According to findings on SPS variables, the two stu- dent populations differ significantly with respect to school aspiration. Although there are observed differences between means for other SPS variables across the two student popu- lations, none of them appeared to be significant. In the next section of this chapter is presented the procedure for developing indices on selected items and assessing the reliability of the data. Developing Indices on Selected Items and Reliability_Analy§is In this section, the procedures for developing some indices and reliability analyses of those indices are explained. Some SES and SP8 variables, used in the data analysis, were developed on the basis of combining or group- ing of subjects' responses to certain items in the Student Questionnaire by means of either simple summation or apply- ing the summated rating scales (also called Likert-type scale).2 Combining the student's response to selected 2F. N. Kerlinger, Foundation of Behavioral Research (New York: Holt, Rinehart and Winston, 1964), pp. 483-491. 61 items for getting a single score (index) was not done on an arbitrary but on a logical basis, that those selected items were assumed to be measuring the same thing by asking stu- dents in different ways in order to eliminate the chance Of error. After the scores on selected items for each index for each student were summed, then internal consistency reliability estimates were obtained by Hoyt's analysis of variance method for all indices in this study.3 The main procedure for each index is explained in succeeding pages. Index of Residence This index represents the sums of scores each sub- ject gets from items 10 to 17 included in the Student Ques- tionnaire. The items indicate whether the student's home does or does not have certain facilities, each of which is assumed to contribute to a higher standard of living. There- fore, it was assumed that the higher the score a student gets on those items the better living condition exists at his home. It was further assumed that it was not necessary to estimate the internal consistency of those items, because of the fact that the existence of any one of these facili- ties at home indicates better living conditions. So, sums of scores on this group of items give us an "index of residence" for each student. 3Hoyt, op. cit., p. 108. 62 Index of Self-Concept of Ability This index was also obtained by summing the scores that students get on items 20 to 26 inclusive, under the assumption that those items meet the requirement of applying summated rating scale procedures. A summated rating scale is a set of attitude items, all of which are considered of approximately equal "attitude value" and to each of which subjects respond with degrees of agreement with the options given under each item in the survey instrument in this study. So, the scores of the items of such a scale are summed to yield an individual's attitude score.4 Thus, the obtained self—concept of ability index was tested against zero correlation among the items used for this particular attitude by means of Hoyt's reliability esti- mate for internal consistency. According to Hoyt's analysis of variance method for estimating the reliability of such an index based on more than one item, the variation in the response of an individual from item to item is not consid- ered to be error at all. Rather, it is a sum of individual differences and residual.5 So, internal consistency-— reliability--can be defined as the ratio of difference between individual variance and error variance to variance of individual. By applying Hoyt's method for internal 4The words "attitude" and “perception" are used as synonyms in this study. 5G. C. Helmstadter, Principles of Psyohological Measurement (New York: Appleton-Century-Crofts, 1964), p. 73. 63 consistency the reliability estimate of self—concept of ability index was found to be .79. Indices of Perceived Evaluation by Others These indices were based on the items which indicate how a child perceives his academic ability with respect to significant others, namely parents, teacher, and friend. Perceived evaluation indices based on groups of items 28-33 (excluding item 31), 35—39, and 41—44 inclusive indicate how a child interprets (perceives) the expectations of his par- ents, best teacher, and best friend concerning his academic potentialities. For developing these three indices, the summated rating scale procedure and Hoyt's analysis of variance method for internal consistency reliability estimation were applied by adopting the same assumptions and logic as we had for self-concept of ability. As indicated in Table 11, the obtained Hoyt's relia- bility seemed to be satisfactory for further treatment by using indices. Correlational analysis is presented in the next section of this chapter. Correlation Analysis The purpose of the correlational analysis was to determine the relation of the academic achievement to (l) socio-economic background of students, and 64 Table ll.-—Internal consistency of indices based on selected items. Number Number of of Hoyt's Standard Indices Items Subjectsa Reliability Error Self-concept of ability 7 742 .79 1.3979 Perceived evaluation by parents 5 735 .80 1.0991 Perceived evaluation by teacher 5 733 .84 1.0380 Perceived evaluation by friend 4 738 .82 .9039 aSubjects who responded to items 28, 35, and 45 as fThey do not care about my education," were dropped from the analysis. (2) socio-psychological factors. More precisely, the pur— pose of the analysis was to provide meaningful answers to the research questions "one“ and "two" stated in Chapter I. The data pertaining to questions one and two are analyzed and discussed under each of these two questions. What is the magnitude of the relationship, if any, between a student's academic achievement and his socio-economic status? In order to provide data for answering this question, the students in the sample were asked to answer the items in the "Student Questionnaire" related to their educational background, father's occupation, father's income, father's education, and home conditions. The data for students' academic achievement were obtained from each student's file at the school. 65 The definitions of socio—economic status variables and academic achievement variables were given in Chapter I. The magnitude of a student's score on SES variables indi- cates the level of his socio-economic status in the popula- tion of interest in this study. In order to test significance of relationships between SES and achievement variables, the above question has been translated into a research hypothesis as follows: Hypothesis 1: There is a positive relationship between the academic achievement of a student and his socio-economic status. The pertinent data were analyzed by using the .CDC 6500 BASTAT program available at Michigan State Univer- sity.6 The correlation matrix based on combined sample (non-Gecekondu and Gecekondu) is presented in Table 12. Each coefficient of correlation was tested against zero correlation at the .01 level of significance.7 All correlations greater than .094 are significant at the .01 level. Therefore, for the combined sample, all correlations are significant, indicating support of the research hypoth- esis except for the correlation of educational background with mathematics. The correlation of achievement variables with socio- economic variables ranges from .03 for correlation of 6Carroll, Donaldson, and Price, op. cit. 7Quin McNemar, Psychological Statistics (New York: John Wiley and Sons, Inc., 1963), pp. 136-168. .Ho. um pamoamacmamy .pmuuHEo mucflom Hmeflomp cam mmomam ABEAOOU osu Op cmpcson amen m>mc mesmHOHmmmoo damn .mDGOOSDmIIspcoxmomw cam sccoxmomwlcoc cuoo mo mcflumHmCOOIlmcmnm prflm was com: comma mum meowumamnuoo onem 66 row «mm «mm gem 4mm «mm 44H muomflnsm m>Hm do .a.m.o .m «om «mm «om «em «om mo mOHumEmnumZ .5 .mm «mm .sm .mm .ma maflemmm .G 3mm 4mm «ms 4nd mucmeummm .m «mm «mm «ma coflumospm m.Hm£umm .w mvnnz «as «am meoocfl m.Hm£umm .m ngma coaummsooo m.umcumm .m ccsoumxomn Hmcoaumoscm .H m h o m v m N a meanmwum> .mHmEmm m pwcwnEOo mo mmHQMHHm> ucmEO>wH£om can mmm chEm cowpmamuuooumucHll.ma magma 67 educational background with mathematics to .34, correlation of father's education with G.P.A. of five subjects, respec- tively. As indicated in Table 12, father's educational level has correlated with student's academic achievement most and father's occupation is the next. After two SES variables, the correlation of achievement with residence and father's income came as correlated SES factors with school achieve- ment. It seems that educational background of a student is the least correlated SES variable with his academic achieve- ment. Further correlation analysis was carried out on the data of the same variables by separate sample (non-Gecekondu vs. Gecekondu students). A summary of intercorrelation is presented in Table 13. The purpose of this analysis was to determine the significance of association between achievement and SES variables when the total sample has been split into two sub-categories-—non-Gecekondu vs. Gecekondu. Table 13 shows that the correlation of coefficients of academic achievement with SES based on the data of the non-Gecekondu sample still holds significant and supports the research hypothesis (positive relationship) except edu- cational background. On the other hand, the analysis based on the data of the Gecekondu sample indicates that only father's income correlated with reading and G.P.A. of five subjects, and educational background correlated with reading 68 can mmomHm Hmfifiomp 03D Op pmpcsou comp m>mz mucmflowmmwoo Had .Hm>m4 40. um namoamaamam. .cmuufleo mucflom Hmeomp a .Hmnommflc came mnu m>onm cm>flm mum mpcmpsum secoxmomo MOM chHpmHouuoo map can .mao mo mmfluucm cues unmau Hmsoa Op umma Home: Eouw mcflccsu .Hmcommac came map soHOQ cm>flm mum mpchSDm socoxmomoncoc How chHumamnnoo crew msmuz 004 .mm «am «mm «64 «mm 444 4mm whomflnsm m>4u do .4.m.o .m 4ms 004 «we «4m 4mm 4cm «mm «NH monumsmnumz .5 «ms .m4 ooa 4mm .m4 4mm .m4 .mm maflemmm .0 mo mo Ho ooa «mm 44s «mm .mm mocmeammm .m 4mmuz so so mo 44m 004 “so «am 4mm coaumosem m.nmnumm .4 «Ha mo «OH «mm «mv OOH ton «vm mEOOcH m.Hm£pmm .m no mo 40 4pm «om 44m 004 4mm aoflpmmsooo m.nmnumm .m 40 so- «04 mo 44H «SA has 004 easonmxomn Hmcoeumosem .4 m A e m 4 m m 4 mmanmflua> m.mHmEmm Spcoxmomw .m> scaoxOOOUIcoc mo moanmflum> unme>mw£om cam mmm macaw :oflpmamnuoouougHil.mH canoe 69 significantly at .01 level. Although, with the exception of the correlation of student's educational background with mathematics, which is negative, all other correlations between SES and achievement were positive, none of them was found to be significant at .01 level. The lower correlations or no correlation at all between SES variables and achievement in Gecekondu popula— tion comparing them with correlations in non—Gecekondu pOpulation indicate that poor students in Gecekondu sample may have as high achievement as those who were a little higher in their socio-economic status and that SES and achievement do not go together in Gecekondu population as well as they do in non-Gecekondu population. One possible explanation of this difference between the findings for Gecekondu and non—Gecekondu populations may be that stu- dents in Gecekondu population may represent more homogeneity with respect to SES variables than do non-Gecekondu popula- tion even though Gecekondu students vary in their achieve— ment. Further explanation will be given in the section entitled "Multiple Regression Analysis." The correlation of academic achievement with combined SES variable is presented in the section entitled "Factor Analysis" in this chapter. The correlational analyses, based on the data pro- vided by the items measuring student's responses to socio- psychological factors in the "Student Questionnaire," reveal the relationship between student's academic achievement 70 and socio-psychological factors. The analysis was carried out once on the data of combined sample (non-Gecekondu and Gecekondu students together) and once on the separate sample. The research question pertaining to this analysis was stated in Chapter I as follows: What is the magnitude of the relationship, if any, between a student's academic achievement and socio- psychological variables? The data presented for this analysis were obtained from the student's responses to the items relevant to socio-psychological factors in the "Student Questionnaire" and achievement scores from his file at the school. The definitions of each SPS variable were given in Chapter I. Except for the school aspiration scale score, the other scale scores are based on more than one item response of student. Therefore, the reader should keep in mind that student's score on self-concept of ability, and score on his perceived evaluation by significant others—-parents, teacher, and friend--indicate combined scores or an index based on more than one item. Development of those indices has been explained in the preceding section. The research question above was translated into a research hypothesis as follows: Hypothesis 2: There is a positive correlation between a student's academic achievement and socio-psychological factors. 71 The computed coefficients of correlation among the variables based on combined data of non—Gecekondu and Gecekondu sample are presented in Table 14. Coefficients in Table 14 were tested against zero correlation. For all cases, the computed coefficients of correlation were found to be significant at the .01 level. The correlations between achievement variables and SP8 variables range from 18 to 52. It seems that student's perception of the evaluation of his academic ability by parents has the highest relationship with student's aca— demic achievement. The next SPS variables which have positive association with academic achievement are self— concept of ability and perceived evaluation by teacher. The least association was found between achievement vari- ables and school aspiration. The analysis also showed that self-concept of ability and perceived evaluation by parents, teacher, and friend have a close relationship with each other. So, this may indicate that those variables are complementary to each other rather than discrete variables. The summary analysis based on the data by school category is presented in Table 15. As indicated in Table 15, all computed coefficients of correlation were found to be significant at the specified level (.01). Although the magnitude of coefficients of correla- tion between achievement and SP8 variables by school cate- gory are not the same, the research hypothesis was supported by the findings. The correlations between achievement and 72 .HO>OH 40. um DGMOHMAcmHm. .pmuuHEo muaflom Hmeflomp Ugo mwomam HmEHomp OBD Ou potency comp o>mz mucmfl0flmmmoo Ha< n .mucmpsumllzpcoxmomw cam secoxmomwucoc OOGHQEOOIIOOOHm 5pxam New com: comma mum mCOflumamHuoo mnem con .04 .mm .km .44 .mm .44 .mm muomflnsm m>am no .<.m.o .4 cos .os .om .mm .44 .mm .44 monuasmaumz .s OOH .mm .04 .m4 .M4 .em manemmm .m 004 .44 .ss .oa .sm ecmanu an coflpmsam>m cm>flmonmm .m N4auz 004 .mm .ms .Hm umnummu an cofiumsHm>m pm>fimoumm .4 ooa «om «mm mucoumm an coaumsam>m pm>amoumm .m can C..em suaaflnm mo ummoaooumamm .m 004 Auammc coflumuaamm Hoonom .4 m h m m v m m a mmHQ6flum> m.mHmEMm pmcflnfioo mo mmaQMHHm> ucmEm>mH£om can mmm mcofim coaumamuuooumucHll.va magma 73 .Hm>m4 40. um ucmOHMACmflm. .Umuuafio mucflom HmEHOOp cam mmomam HmEHomp 030 cu OOOCDOA comb m>mn mDGOHOHHMOOO Hadn .Hmcommflp came map m>onm cm>4m mum wucwosum upcoxmomw How mcoflumamuuoo may cam .wco wo mwfluucm £043 0:044 Hmzoa Op puma Home: Eonm mcflccdu .HmcomMHp same man 304mb qm>flm who mucmcsum occoxmomwlcoc 40m chHDOHOHHOO crew 404uz 004 .44 .04 .44 .04 .44 .04 .44 muom4nsm m>44 mo .a.4.o .4 «mm OOH «mo «Hm «he «om «mv «Hm mUHumEmnumE .b .40 .04 004 .44 .04 .44 .44 .44 4c4emmm .4 .44 .04 .44 004 .00 .40 .44 .44 6co4uu sh coflumsam>m pm>wooumm .m .04 .04 .44 .44 004 .44 .40 .04 4406444 4n ammuz coflumsam>m Um>4moumm .4 smm smm tom «an «vm OOH «Nw «Hm mucmumm >Q coflum94m>m cm>flmoumm .m .04 .44 .44 .40 .40 .40 004 .44 4444444 40 unmoaoo-44mm .4 .44 .44 .04 .44 .44 .04 p.44 004 c044444444 400:64 .4 4 0 4 m 4 4 4 4 464nm4um> m.mamfimm Spcoxmowo .m> Spcoxmom0|coc mo meanmflum> uCOEw>m4£om new mmm mCOEm coflpmawnuoonmucHll.m4 canoe 74 SP8 variables, based on Gecekondu sample, range from .13 to .40 while they are between .21 and .57 for the data of non- Gecekondu. Self-concept of ability and perceived evalua- tion by parents and by teacher seemed to have more associa- tion with academic achievement of student than the association of achievement with school aspiration and perceived evalua— tion by friend in both sub-samples. Generally, the rela- tionship between achievement and SP5 variables is higher for the non-Gecekondu sample than the Gecekondu sample. Why is this so? Probably, the students who have lower grades on reading, mathematics, and G.P.A. of five subjects think they can do as well as others who have had better grades on the same academic area in Gecekondu pOpulation. Therefore, their responses to the items which were assumed to be socio- psychological variables were almost as high as the responses of students who were more successful at school work. They may feel that they have not done as well in their actual school achievement as measured by grades as they think they can do in comparison with other students. Most of the stu- dents in Gecekondu pOpulation might be told by their parents and others that they could do school work as well as others, although many of those parents may not be as sophisticated about school tasks. This suggests that the variance in academic achievement of non-Gecekondu pupil can be more accurately explained than of Gecekondu pupil by means of socio-psychological variables. In fact, in both samples SPS variables seem to explain more variation in academic 75 achievement than do SES variables. The relationships among Vsome selected variables, including those above, in the com- bined sample as well as in samples of non-Gecekondu and Gecekondu students are given in Appendix C. Further correlation analysis will be carried out in the next section to see the relation of composite measures of SES with achievement and of composite measure of SPS with achievement on the data of non-Gecekondu vs. Gecekondu stu- dents and on the data of the combined sample. Factor Analysis for Developing SES and SP8 Indices The purpose of factor analysis in this study is to search for variables--items or indices--that correlate substantially with one another so that they can be grouped as new indices which can not only serve for further analy- sis but are also psychologically and sociologically meaning- ful. In the development of the "Student Questionnaire," some of the items were so constructed that one group of items (#4, #7-17) was expected to provide information about student's socio-economic status (SES) and the other group of items (#18, #20-26, #28-33, #35-39, #41-44) was related to the student's school aspiration, his perception of himself and evaluation by significant others with respect to his schooling. The procedure for developing indices of residence, self-concept of ability, perceived evaluation by parents, 76 perceived evaluation by teacher, and perceived evaluation by friend has been explained in section two of this chapter. Now, by applying factor analysis to data on those selected variables it was intended to identify each variable whether belonging to the SES group or to the SPS group as they were constructed. Therefore, the intercorrelations of those selected variables were subjected to a two-step factor analysis known as principal components (or principal axes) analysis and varimax rotation analysis in the literature.8 These techniques are part of a broad family of techniques generally known as factor analysis. The principal components method is used quite often as the first step in a step-wise analysis. This method is the ideal method of condensing (grouping) variables during the first step of a two—step analysis. In principal com— ponents analysis, the first principal axis is defined as that linear combination of variables which explains the most variance. That is, weights (or factor loadings) for the first factor9 (component) are selected so that the average squared factor loading is maximum. Then the first 10 residual matrix is obtained. A linear combination is then formed of the partialed variables in the residual matrix, 8P. Horst, Factor Analysis of Data Matrices (New York: Holt, Rinehart and Winston, 1965). 9"Factor" and "component" are used synonymously in this study. 10J. D. Nunnally, Psychometric Theogy (New York: McGraw-Hill, 1967): pp. 288-347. 77 so that the average squared loading on the second factor is as large as possible. This procedure is repeated until the desired number of factors is extracted. In the second step of factor analysis, applied to data in this study, the varimax rotation technique has been used in order to get simple structure; that is, to identify variables having high factor loading on one factor (being important to that factor) and of negligible importance to all other factors. In short, this technique attempts to maximize the high and low weights for a factor so that the variables that have high weights (factor loadings) on a factor can be thought of as belonging together. In this way, an interpretative label can be applied to what they have in common. The following part of this section describes the application of principal components and varimax rotation techniques to data of this study for develOping SES and SP8 indices. One of the important objectives in this study was to see in what manner responses to different items in the Questionnaire related to one another. In this way it was hOped to see linear relationships of variables among them- selves which could be explainable by student's socio-economic status or by socio-psychological factors. Each of them can also be used as an independent single composite measure for further analysis to predict student academic achievement on selected subject areas. 78 Intercorrelations of ten variables, subjected to two-step factor analysis, for combined pOpulation (non- Gecekondu and Gecekondu) are given in Table 16. As indicated in Table 16, the coefficients of the first five variables, which are measures of items which were constructed for get- ting information about student's socio-economic background, range from 17 to 85. The correlation of student's educa- tional background with the other four SES variables is not as high as the correlations of four variables among them— selves. However, the correlations of educational background with SP5 variables are far lower than their relationship with SES variables. Therefore, not only because of the magnitude of the coefficient of correlations, but also by definition, student's educational background variable can be treated as one of the SES variables rather than one of the SPS variables. Father's occupation, father's income, father's education, and residence correlate with each other high enough that, to a large extent, they may be assumed to be measuring a common attribute. The same interpretation can be applicable to the variables based on items which were constructed so that they measure the level of student's responses to socio-psychological factors. The range of coefficients of correlation based on SPS variables-~school aspiration, self-concept of ability, perceived evaluation by others (parents, teacher, and friend)--range from .31 to .83. The school aspiration variable correlation with other SPS variables is not as high as that of the other four SPS 79 popcsou coma m>mn m054m> 444 .pmuuflfio mucHom Hmefiomp paw moccam Hmeflomp 034 on .mucmpsum ~40 com: comma OHM mCOHn—MHOHHOUM 004 44 00 00 04 04 44 04 04 40 444444 04 GOHUMDHM>O U®>HOOHO4H .OH. 44 004 44 40 44 04 44 04 44 40 4444444 04 c0444544>m Um>flmoumm .m 00 44 004 04 04 44 44 44 04 40 4444444 44 c0444544>m pm>flmoumm .m 00 40 04 004 04 04 44 44 04 00 4444444 44 444oaooum444 .0 04 44 04 04 004 44 04 44 04 40 4444444444 444440 .0 04 04 44 04 44 004 40 00 40 04 444444444 .0 44 44 44 44 04 40 004 40 04 04 444444444 4.444444 .4 04 04 44 44 44 00 40 004 40 44 444444 4.444444 .4 04 04 04 04 04 40 04 40 004 00 4444444444 4.444444 .4 40 40 40 00 40 04 04 44 04 004 0444444444 44444444444 .4 04 0 4 0 0 0 4 4 4 4 44444444> M .m44n4444> mmw can mmm pmuomamm mo xflnume coflumamuuooumucHll.m4 manna 80 variables with each other. Nevertheless, by definition it refers to student's attitude rather than to his socio— economic background. Following the interpretation of those coefficients of correlation in Table 16, they were then subjected to two-stepwise factor analysis, as described previously, by using the CDC 6500 CISSR FACTORA program11 for condensing (grouping) the variables as single indices of SES and SP8. The figures in Table 17 indicate the factor load— ings12 (correlations of the variables with factor scores) of each variable on a particular factor. For example, the factor loadings of educational background variable are .19 with the first factor, .26 with the second, and .92 with the third factor, and so on. Father's occupation variable has factor loading .68 with the first factor, .60 with the second, and -.07 with the third factor. As indicated in Table 17, almost all of the vari- ables have highest factor loading on the first factor. That means the first factor explains the most variation, 43 per cent, in achievement. Regardless of the sign of factor loading, next to the first factor, the second factor seems to explain a considerable amount of variation, 24 per cent, in student achievement. The signs of the factor loadings on the second factor indicate that the first five and 11Price and Ingvaldson, Op. cit. 12Nunnally, Op. cit., p. 292. 81 .04Q4444> £040 co mGHUmoa Houomm musmmmummn 045m4m comma .pmuuflfio musflom 4484000 0cm mmomam 4484044 0344~ Ou UmUGSOH 4.4me m>mn mwflam> Had .mucmmvflum NVB fiOQD meMQ m..n mHm%HMC¢m 444 44 44 44 44 44 44 44 44 44 44444444 44 4444 444 4>4444444o 44 44 4o 44 44 4o 44 44 44 44 44444444444 444444 4444 44 444 pouc5000m 00s4444> mo 4:00 404 44 4o 44- 44 44 4o- 44. 4o 44- 44 444444 44 20444944>0 ©m>400umm .04 44- 44- 44 44 44. 4o- 44- 4o 44- 44 4444444 44 GOHHVMDHMNVT Um>HmOme .m 44 44 44 4o- 44. oo 44- 4o 44- 44 4444444 44 c0444344>0 pm>flm0umm .m 44. 44- 44- 44- 44- 4o 44. 4o 44- 44 4444444 44 4444444-4444 .4 44 4o 44. 4o 44- 44- 4o 44- 4o- 44 4444444444 444444 .4 44. 4o- 44 44 4o 44 44 44- 44 44 444444444 .4 44- 44 4o- 44 44- 44 44- 4o- 44 44 444444444 4.444444 .4 44. oo 44 44- 44 44- 44. 4o- 44 44 444444 4.444444 .4 44 44- 44 4o 44. 44 44- 4o- 44 44 4444444444 4.444444 .4 4o 44 44 44 44. 4o 44 44 44 44 4444444444 44444444444 .4 04 m m h m m e m N a 40434444> m m 0 B 0 d h m .44494444> mmm can mmm £443 4404044 cmumuoncs mo xHHumE mcapmoa Houommuu.ha manna 82 the next five variables clustered in different domains of factor space. Since the purpose of the factor analysis was to find out simple structure, i.e., that SES variables have high fac- tor loadings on one factor and SP3 variables have high factor loadings on the other factor, varimax rotation technique was applied to principal components matrix. The results of rotated factor loadings are presented in Table 18. The results of rotated factor loadings for the first two factors are presented in Table 18. The remaining factors were not included because theory had suggested explanation of performance in terms of two factors and because the factor loading matrix of Table 17 gives empirical evidence in sup- port of two major factors explaining the majority of the variance (67 per cent) in student achievement. As indicated in Table 18, SP5 variables--school aspiration, self—concept of ability, perceived evaluation by others (parents, teacher, and friend)--have high factor load- ing (weight) on the first factor and SES variables--educational background, father's occupation, father's income, father's education, and residence--have high factor loading on the second factor. School aspiration has almost equal weight on both factors; by its definition it was assumed to be an SPS variable. However, its contributions to the magnitudes of condensed SP8 or SES variables do not differ very much and its inclusion for computing student's score on SP8 and on SES do not affect further analysis. The same reasoning can be made about educational background factor. Table 18.--Rotated factor loadings on two orthogonal factors.a Factor Loadings Variables Faitor Fagtor Commufiglity 1. Educational background -02 32 10 2. Father's occupation 14 89 81 3. Father's income 12 86 75 4. Father's education 18 88 80 5. Residence 13 81 67 6. School aSpiration 44 32 30 7. Self-concept of ability 88 08 78 8. Perceived evaluation by parents 92 13 86 9. Perceived evaluation by teacher 91 05 84 10. Perceived evaluation by friend 90 05 80 Per cent of variance accounted for by each factor 35 32 Cumulative per cent of variance 35 67 aAnalysis is based upon 742 students. All values have been rounded to two decimal places and decimal points omitted. Figures in the third column of Table 18 show the "communality of the variables," which is the sum of the 84 squares of the common-factor coefficients.l3 Communality indicates the extent to which the common factors account for the total variance of the variable and the remainder is called uniqueness. The main purpose of this two—step factor analysis was to develOp a single index for SES and a single index for SP8 for each student. In order to get student's SES index, each student's scores on ten variables were multiplied by their factor loadings on second factor and then summed as follows: SES = 32(ed. back.) + 89(f. occup.) + 86(f. income) + 88(f.ed.) + 81(residence) + 32(school asp.) + 08(self—concept) + 13(sig. par.) + 05(sig. teach.) + 05(sig. friend) In order to get student's SPS index, the same compu- tation was carried out by substituting factor loadings on the first factor in the formula above. In this way, two indices, as standard scores, for each student were created and punched on an IBM card with the student's identification by using CDC 6500 FACTORA-PUNCH program available at the Computer Center of Michigan State University. The next section deals with the application of regres- sion analysis on the data--SES and SP8 indices in standard score form--in order to provide answers to questions 4a, 4b, and 4c proposed in Chapter I of this study. 13Harry H. Harman, Modern Factor Analysis (Chicago: The University of Chicago Press, 1960), pp. 14-16. 85 Multiple Regression Analysis This entire study is based on the assumption that at least some of the differences in academic achievement of sixth grade students in Ankara can be explained by the dif- ferences in social-psychological factors and by the differ- ences in their socio-economic status. The objective, then, is to find out the extent to which each characteristic correlates with the student's academic achievement on reading, mathematics, and G.P.A. of five subjects, inclusive of reading and mathe- matics. Therefore, the student's score on his socio-economic status measure and his score on his attitude measure (inde- pendent variables), and his grades on subject areas (depen- dent variables) specified above were subjected to multiple regression analysis by using the CDC 6500 LS program.14 Stated broadly, the purpose of the multiple regres- sion technique (sometimes called multiple prediction) is the estimation of a variable y (dependent), from a linear com- bination of m independent variables which may be identified as x1, x2, . . ., xm.15 When the predictor variables are statistically independent, multiple regression provides information about the relative importance of the predictor variables for the explanation of variance in a dependent (pre- dicted) variable. In summary, it can be said that the basic l4Carroll, Donaldson, and Price, Op. cit. 15G. V. Glass and J. C. Stanley, Statistical Methods in Education and Psychology (Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1876), pp. 186-191. 86 objective of application of multiple regression analysis to data is to obtain the Optimum weighting (Optimum con- tribution) to be assigned to each independent variable in predicting a dependent variable.16 This section indicates by means Of regression anal— ysis how much variation in the academic achievement of a sixth grade student included in this study can be accounted for by his measured socio-economic status and by his measured perception of himself and others' evaluation for achieve- ment. The section also includes the results of regression analysis based on the data of non-Gecekondu and Gecekondu schools separately, in order to see whether the student's socio-economic background and socio-psychological factors function differently in the two groups in predicting the student's academic achievement or in explaining the variance in academic achievement. It may be recalled that the procedure for develOp- ing an index—-called SPS—~based on variables representing the student's school aspiration, his self—concept of ability, and perceived evaluation by others (parents, teacher, and friend), and an index--called SES-—based on variables rep- resenting the student's educational background, his father's occupation, his father's income, his father's education, and his residence condition, has been explained in a pre— ceding section. SES and SPS indices are in standard score l6McNemar, Op. cit., Chapter 11. 87 form and punched on an IBM card for each student with his scores on specified subject areas representing his academic achievement. Each card has also a code indicating whether a student belongs to non—Gecekondu schools or to Gecekondu schools. Before evaluating the results of regression analysis based on the data of combined sample (students of non- Gecekondu and Gecekondu schools together), it may be instruc- tive to look at the correlations of SES and SP8 indices (also called independent variables or regressors or predictors) with each of the academic measures called dependent variables (or criterion variables or predicted variables) of interest in this study. The intercorrelations are given in Table 19. Table l9.--Intercorrelations of the predictors and dependent variables based on combined sample.a Variables l 2 3 4 5 1. SP8 index 100 2. SES index 00 100 3. Reading 41 29 100 4. Mathematics 38 25 60 100 5. G.P.A. of five subjects 48 28 83 80 100 aCorrelations are based upon 742 students. All values have been rounded to two decimal places and decimal points omitted. All correlation coefficients were found to be significant at .01 level. As may be seen by inspection of Table 19, the cor- relation between SPS and SES indices is zero. In fact, this 88 is not an accidental outcome. It may be recalled that each of those indices represents a composite measure on a factor. In other words, SPS index is the composite measure based on the first factor and SES index is the composite measure based on the second factor, as was explained in the preceding sec- tion. Since the first and second factors are orthogonal (independent from each other) the correlation between SPS index and SES index has to be zero., Further inspection Of Table 19 reveals that the correlation of SPS index with dependent variables-—reading, mathematics, and G.P.A. of five subjects-~13 always higher than the correlation of SES index with the same dependent variables. This indicates that the differences of academic achievement of the students can be explained better by the differences Of socio-psychological factors than by the differences of their socio-economic status. In order to provide a statistically meaningful answer to the question of which index is more closely related to the observed differences in academic achievement, SP8 and SES indices with dependent variables (achievement variable) were subjected to multiple regression analysis. The beta weights (relative contribution of each index) and multiple correlations for estimation of each achievement score of the student are summarized in Table 20. Inspection of Table 20 indicates that SPS has beta weights which are always higher than the beta weights on SES. The difference between two beta weights for each dependent 89 Table 20.--Beta weights and multiple correlation for the SPS and SES indices in estimation Of academic achievement.a Zero Order Achievement Multiple SPS & Variables Beta Weights Correlation Achievement . 2 2i (Predicted) N SPS SES R1.23 R1.23 r r Reading 742 67 47 50 25 41 17 Mathematics 742 75 49 46 21 38 14 G.P.A. of five subjects 742 76 45 56 31 48 23 aAll values have been rounded to two decimal places and decimal points omitted. All multiple and zero order correlations are significant at .01 level. variable is so large that SPS alone seems to be a simple predictor to estimate the student's grade on those subjects. But in spite of the lesser weights on SES which were found for each achievement variable (dependent or predicted vari- able) as compared with SPS, the contribution of SES to pre— diction of dependent variable is still substantial. When the square of multiple correlation and Of zero order correlation was computed it was found that the addition of SES variable to SPS variable in regression equation increased the explana- tory function of SP8 on each criterion (dependent variable) some 32 per cent on reading, 33 per cent on mathematics, and 26 per cent on G.P.A. of five subjects. This means that for prediction of student's score on those subject areas with addition of his SES score to his SPS score the predictability function of the regression equation is increased. Thus SES 90 variables as a composite index can be used as an additional independent variable for the prediction of student achieve- ment instead of using SPS alone. As stated in Chapter I, one of the objectives of this study was to investigate socio-economic and socio- psychological factors (represented as SES and SP5 indices, resPectively) across the population in order to see whether those factors have a different pattern of relationship with student's academic achievement or not. Table 21 shows the simple correlation among selected variables of interest for Gecekondu and non-Gecekondu students. Table 21.--Intercorrelation among selected variables (non-Gecekondu vs. Gecekondu).a Variables l 2 3 4 5 1. 593 loob -2s 36 33 39 2. SES 17° 100 oo -03 01 3. Reading 48 39 100 49 75 4. Mathematics 45 35 66 100 75 5. G.P.A. Of five subjects 58 40 87 82 100 aCorrelations below the diagonal are based on non— Gecekondu data (N=378), and those above are based on Gecekondu data (N=364). bAll correlations of coefficients above 094 are sig— nificant at .01 level against zero correlation. CAll correlations have been rounded to two decimal places and decimal points omitted. 91 Table 21 shows the intercorrelations among variables based on two separate samples as compared with Table 19 based on the combined sample. In Table 21, the correlations below the diagonal are for non-Gecekondu and those above for Gece- kondu sample. Investigation of Table 21 indicates that the two pop- ulations differ not only by Observed differences in magnitude of correlations of coefficients Of SPS with achievement and SES with achievement but also by patterns of coefficients. Plausible explanations for these observed differences between the two pOpulations are as follows: The correlations of coefficients based on non— Gecekondu for SPS with achievement were found to be higher than the correlations of coefficients based on Gecekondu sample. The low achievers in Gecekondu might be motivated by parents or others with whom they interact in such a way that education is exceptionally important for them compared with other things in their life. The low achievers may see education as the only means for having a better life or step- ping up the social ladder in a broader society. So, they may have an unrealistically high self-concept, high perception about evaluation of their ability by others, and probably high school aspiration, as much as those who seem to be successful in the same academic area. The most interesting thing that Table 21 reveals is that SES did not correlate with any Of the achievement variables for Gecekondu data, while correlation is substantial 92 for non-Gecekondu. Before giving an explanation for this finding, it may be interesting to explain why SPS and SES correlated negatively for Gecekondu population, whereas the correlations in Table 21 were based on separate sample data. In other words, the analysis procedure also reflects the trancation on SES which has occurred by separation of com- bined group into a Gecekondu (assumed to be primarily low SES) and non-Gecekondu (assumed to be primarily high SES) sub-samples. Splitting the combined group into two sub- samples decreased the heterogeneity which generally goes with the magnitude of correlation of coefficient. Further analysis was then carried out to see whether Observed dif- ferences in the two populations are significant or not. The analysis of the data based on sub-samples (non- Gecekondu vs. Gecekondu) was carried out by means of regres— sion analysis to see whether the SPS and SES indices (independent variables) are functioning in the same manner in the equation for predicting student's achievement score on each of the selected academic subjects. Thus, the regression analysis was conducted on data of each sub—sample separately in order to provide answers to questions 4a, 4b, and 4c posed in Chapter I Of this study. The questions concern whether student's socio-economic back— ground factors (composite SES index) and socio-psychological factors (composite SPS index) play different roles in pre- dicting student's achievement score on specified subjects across the two samples. The difference between regression 93 equations across the samples was tested for significance at .01 level. In this line, each question (4a, 4b, and 4c) has been translated into a testable research hypothesis and tested as follows. Question 4a: To what extent, if any, do SES and SPS variables differ, between the two student pOpulations, in predicting students' achievement in reading? Research Hypothesis 3: Beta weights (relative contribu- tion) of SPS and SES indices in predicting of student's achievement on READING differ by student populations. Regression analysis was conducted once on non- Gecekondu data and once on Gecekondu data with the dependent variable of reading. Then the differences between the beta weights of the two equations were tested at .01 level of sig- nificance. The summary of analysis is given in Table 22. Table 22.--ANOVA table for testing the equality of regression equation with SP8 and SES indices (independent variables) for predicting READING on two samples at .01 level. ‘ --7.“-_.- Sums Of Source of Variance d.f. Square F Deviation from hypothesis (null hypothesis) 2 71.353 18.437 Sig. Separate regressions (residual) 736 1420.615 Common regression 738 1491.968 Inspection of Table 22 reveals that computed F was found to be significant and suggests that the relative weights 94 of SPS and SES for predicting student's grade on reading differ by student's population. Therefore, the research hypothesis is supported with findings. After having found the F value significant, post hoc techniques will be used to test the pair Of beta weights in the regression equation to see which independent variables produced the differences between the two populations in pre- dicting student's grade on reading. It may be useful to mention that the use of post hoc technique to test the dif- ference of beta weights is a different process from testing the difference between two regression equations based on two separate samples. It will be remembered from the explana- tions given at the beginning of this section that the regres- sion technique is the estimation of a variable Y (dependent) from a linear combination of m independent variables X1, X2, . . . , Xm' The expression of regression equation based on Gecekondu sample and non—Gecekondu sample in this study can be presented in formula as follows: Yo: a + BSPS XSPS + BSES XSES _ 1 1 1 1 1 YNG ‘ a + Bsps Xsps + BSES XSES Where G stands for Gecekondu and GN for non-Gecekondu, and a is the constant or intercept. The tests by using post hoc _ l _ 1 8SP5 ‘ 8SP3 and 8SES ‘ BSES for the two sub-samples. The summary analysis is presented were only for: simultaneously in Table 23. 95 Table 23.--Testing the differences Of beta weights across the sample for each independent variable (SPS and SES) with READING at .01 level. ——-——- ‘__- _—_-.. . Confidence Variables Beta Weights Interval (Indices) Non-Gecekondu Gecekondu dif. ona :0>«o on: ounces». acooxoooo new ucouuoaouuoo on» use .ooo no coauuco Audi unuuu aura” o» awed “one: song oceans» .«qcououv cans oz» radon co>am on: Queens». :ucoxooooucoa new ocoauaaouuoo Isa- IflOOnAflI nvhlz ooa on an vv vH no on no on no no on on no on an Auaoauuo coauuuamuc Aoonoo oo>aoouom .Hn no ooa no no Ad «on oa or no no on on no ov ov ov guano-duo noduquaa-I deacon uo>uoouom .oa on ow ooa on ma «on ma Ad on on no no on so on ov ..ucoudao caducuamoc goose. vo>woouom .oa no on an ooa no no ma oo an a“ an no on an «a do .oucovsu- ocean. coauuuaaoo «cone. oo>qoouom .oa no «a ma oo ooa van ha AA na ma .4 «a «a be ma on 00¢ .oa man was no: «a van ooa oat ca: no: no: mo: no- no: no: ma: on: New .oa oa me No no NA an: ooa ao oa on ma no «a on mm on Raccoon» ocoeuv oudfluao owaovqud .na oa va Ad ad oo oat AH ooa «on '0 mo ca «4 ma «a ma Anaconauo 0:061. ouuaaau caaovnud .va an on on ma oo can ma no: ooa or no on an av an on occuuu an .anbo u0>aoouom .nu om cm on no no HNI ad val we cod no as on hm 6v hv M05000» ha .HIDO ul>dioula .NH on on on ma on can do no or on ooa no an No on an cascade an .~u>o oo>uouuom .HA am an Nm o« no MAI nu do up vs no ooa mm on nv an hudaand no umOOQOOIuHom .oa on on on «v «A «on ao oo do on ov av ooH on an an and... nodunuaaud Hoonum .o A.:uua can unavnuu .0:«. on an on so: do no: mo ca an on on co co ooa no no nuuofins- o>au «o .1.m.o .o on on on we: no no: no no on on on nu ma no cod on uuaudaonuq: .5 on on on vo: vo own «a mg on an on on ad no no ooa nuances .o no «a oa no «o Ad: vo no- vo oo oo oa «a no no Ho ooaovauou .n oa mo oa no: «a AH: no no: no no no AA «a oo no no cauuauauo -.uonuah .v ea Ha ma mo ca can oo «o no no no mo mg a“ no o4 canon“ -.uonu~h .n oo oo oo «o oa von no oo do: no «o «on oo oo oo vo noduqaaooo o.u¢£udh .u co we: no: no: no out we «on «o .0 no: not do vo no on o unnoumxonn unscaudoaun .4 )I, )ll ) ) ) A )3 H11 as... Ls. as. me... u a. ms on 4" mafia-..“ ...». an. as m .... a an nu 3 an Jun: aui uui ui .- x In. MTV 1 1 at: '91 I au. DWV .4 I I n3 3.4 3.4 on T. a w. o 1. O . 0 Omar may 1110 97.3 1.13 W.) T D?» W D. t. D H q n u 1.0 a a a a a a u o u a u a one one ama % 3m .30 a I D. no me do 59 w”... mu... “w... 5w... .3 ......m We... wan ...... no (I .3, m . . .. .. .... .. .kua .109 ears clue I q. S .u 7‘13 11:9 n.+a .:u V ”(a 3 6 w Mum. h W.n 0.” .P (.P (Lv3.P o 3 op (o? (o? no a I I a u o wu m m m u u u R“ d W)A a u Pu H~ om oa oa ha ma mu va ma «a 44 ca o o o o n v n a A a.uamx DOZOIflUmUIzoz no ‘910 HEB IO num‘n mflAn¢H¢<> awhbfldmm ho zOHH¢AudmOO¢flBZH 163 .oouunfio nuance nufinuov van nooonn «454000 03» cu oovcnou coon o>un uncanoauuoouq IUOOHASO n vhlz can on nn nn me no: mg on on mm on an on no on on on An on nu no Annunuu. canuaunm-n oo>nouuom .An ooa an an ma no: on an on on on nn nn no nn on on on on on no .uonucuuo noduauaaoa 00>«oouom .on oon on an no: no an on an on nn no on nn an on nn an nn no .uunouua. canvuunan- vo>noouom .oa ooa on no no on on na nu an we no an an nn on n« on oo Reasons». wooed. noduounm-a vo>nouuom .on ooa on: on no on on na no an an an ma on an mn on on 00‘ .na oon on: «an no: an: nan on: no: on- no: on: man can but man man now .on can no on on nn on no on an an no nn no nn no Anonuuou ocean. ouuaaao odavoao< .mn ooa no: oo vo no on na on ma an no no on no Anacooauo moose. ouuannu onaovuo< .vn oon no no on on nn on an an an on on no vnomuu an .Ho>o vo>noouoo .na con no no An no on ov on an nn an no Rococo» an .Aa>o vo>aoouom .NH ooa on on an no no mm on an on no mucouda an .nn>o vo>aouuom .an oon on no on nv fin nn on on no hunannd no umoocoonunom .oa oon nn on on an on on on no .uaooo canvaunmuu Hoonom .n ooa on no on on on nn on A.nuna van Ugandan ocwoaaoano .uounnsn o>nu uo .<.a.u .o ooa on on on on on no noduuaonuux .h can on nn pm an on ocnoaom .o oon no no no ha uneven-om .m ooa no no on conuaosvo u.uonu¢h .v ooa an an 0300:“ ..uonuom .n ooa an canuuanouo u.uonuah .n coon vascumxunn aucOAuqoaom .a as. ...fi. “3. m3. .... .1. mm... Mrs ..H .3. 2...... we .3 w... . . . p". u... 3 .m 1.u.1 nu... Jui oui a x In In It... as... in: ... ou. qu .4 a s n: 3.4 a? on a. o a. o a. o u. o u mnr namnr ..To 9 Ta. 1 To «"3 To . “.0 W W I «ou. o H n H N u a H a u o 5 a 6 a 5 m a.n9 a n.u a na . .00 a J u W e.u m.. Ana 6 u PWT 9V3 3V! VT 2m 3 unt H91 uwI ID I 51 m 6 3M aJ v: 13 ( A 16A 38A SBA Lat. Sat. Pan 33A 13.... To u T .4- OT do (do (63 ads 3 a 3 a (re JFa eta Iu v as 3 a o: I In no .P .P .P n.P a n 0? (OP (oP 35 s H.1 I a u o uu . o . u u u K: d vmn a U P: 1‘ u. I.\ d a n4 3 s I . .4 M an on on on ed on ma va n4 nn an on n o h o n v n n a .Ndmtdm DQZOxuUNU 02‘ DQZOKwUmUIZOZ mo nukbmqmm m0 ZOH84Am¢¢OOMNFZH "I71111111111'11111“