Zambezia (1988), XV (i).RESEARCH REPORTSCHOOL TYPE AND ACADEMIC ACHIEVEMENT: 1985JOSEPHINE JORDANDepartment of Psychology, University of ZimbabwePSYCHOLOGICAL RESEARCH OFTEN requires a sample frame representing thediversity of social conditions found in Zimbabwe. In theoretical research, we wish toascertain that a model of behaviour accounts for human responses equally well inall circumstances. In applied research, we evaluate the effectiveness of a techniquein the full range of conditions under which it will be used. For psychometric andeducational research, school type is a possible operationalization of the range ofcircumstances under which Zimbabweans grow up and develop.In 1964 and 1969, Irvine presented evidence on the relationship betweenschool type, school achievement and psychological test performance. In a study ofForm I pupils, Irvine (1964) showed that school types, defined as rural boarding,urban day and rural day, were a more dominant influence in both school andpsychological test performance than the Standard Six barrier examinations.Irvine (1969, 284) classified the schools into three grades: A, B, and C.Grade A Schools Š Single-sex, denominational boarding schools built on CrownLand with long-serving staff, offering four-year courses.Grade B Schools Š Schools, mainly boarding, sharing some of the characteristics ofthe A schools, but including non-denominational government schools and schools in smallurban centres.Grade C Schools Š Schools either in large urban areas or on European farm land,non-boarding and mixed-sex with a high proportion of staff with short terms of service inthese schools.There was a strong relationship between school type and the Junior Certificateresults of 1965.From a wider sociological study in 1971, Dorsey (1975) presentedconfirming and extending evidence of the relationship between school type andexamination achievement. Dorsey related 'O'-level achievement to severalcharacteristics of the schools: size of enrolment; teacher-pupil ratio; qualificationsof staff; racial composition of staff; ethnic composition of student body; socialcomposition of student body; denominational affiliation of school; sex compo-sition of school; and rural boarding-school environment. 'O'-level achievement ofschools was represented by the percentage of pupils who obtained first-divisionpasses. 'O'-level achievement was significantly related to four school character-istics: racial composition of staff; denominational affiliation of school; sexcomposition of school; and rural boarding-school environment. In addition,Dorsey showed that of the top ten schools (in 1971 there were 69 schools offering'O'-level, 67 being included in Dorsey's sample), eight schools were single-sexmission boarding-schools, i.e. schools in Irvine's Group A category.Fifteen years have now passed since Dorsey's data were collected, with4344 SCHOOL TYPE & ACADEMIC ACHIEVEMENTnational independence, the integration of the Ministries of African, European andAsian and Coloured Education, and the rapid expansion of the school system inthe intervening period. This article reports a re-examination of the relationshipbetween school type and examination achievement using the 'O'-level results ofthe integrated and expanded school system of 1985.METHODSchool categorizationIn 1985,779 centres presented 'O'-level candidates. Scrutiny of the records of theExaminations Branch of the Ministry of Education showed that summarystatistics of 'O'-level results are presented by the University of Cambridge LocalExaminations Syndicate.In the light of Irvine's and Dorsey's results the most valuable printout related'O'-level results to school type. The computer output classified schools into eightcategories: Government Group A; Government Group B; Commercial Colleges;District Council; Farm and Mine; Mission; Private; and Rural Council orProvincial Authority. Students who entered the examinations privately werecategorized separately.*A second printout that was relevant to this study compared the performanceof boarding and non-boarding pupils. However, examination of the records of thePlanning Section of the Ministry of Education showed that in 1986 only 12.3percent of pupils were boarders. The low percentage of boarders renders acomplete two-way analysis, boarding by school type, tenuous in parts;consequently, despite the high heuristic value of the boarding/non-boardingclassification, this dimension was abandoned for this study.School type was categorized, therefore, by the eight school types listed above,which are used in the routine statistics presented by the Ministry of Education.Examination classificationIn 1965, Irvine used Form I results, in 1969 Irvine used Junior Certificate results,and in 1971 Dorsey used first-division passes at 'O'-level to represent academicsuccess. This study is concerned only with 'O'-level examination results which areno longer reported by the University of Cambridge Local ExaminationsSyndicate in the form of divisions. The printouts of examination results by schooltype were provided by examination subject (for example, English Language) andlisted the number of children in each school type who attempted the subject andthe number who obtained A, B, C, D, E and U grades. From these data, only twostatistics were used: the percentage of children who obtained A-grade passes andthe percentage of children who obtained overall passes, i.e. the total percentage ofchildren obtaining A, B and C grades.ANALYSESComparison of academic success in subjects offered by all school typesThe academic success of the eight school types was compared in two ways: firstly,'Schools were classified in 1985 by 'Responsible Authority*. The categoi ies listed in the articleare self-explanatory. If an experiment required a more precise sampling frame than the one devisedhere, it would be possible to request further details from the Ministry of Education. The list of schoolsand their classifications for 1985 are available from the author.JOSEPHINE JORDAN 45with the proportions of students who obtained A grades: and, secondly, with theproportions of students who obtained overall passes. Two separate 8x2 chi-squares were calculated to summarize the comparisons and were repeated foreach of the fifteen subjects. For the sake of completeness, chi-squares werecalculated even where very small numbers of students were expected to obtain Agrades or overall passes. Where the data violated chi-square assumptions, it isclearly marked in the results.Rank order of the eight school types by academic successThe 30 chi-square statistics generated in the first analysis indicate whether or notschool type effects examination results. The statistics do not, however, tell uswhich of the school categories obtain higher proportions of A grades or overallpasses, or whether differences in school types are consistent over subjects. Toprovide a rank ordering of school types over different school subjects, the tensubjects which were offered by all school types and which were taken by an excessof 25,000 candidates were employed in a Friedman ranked analysis of variance(Siegel, 1956). The analysis ranks the performance of the schools on each of thesubjects included and computes a summative statistic with the distribution ofchi-square.Comparison of the academic success of school types for subject groupsThe last analysis was prompted by the possibility that some school types may dobetter in mainstream subjects and others may do better in alternatives. All theexamination subjects taken in 1985 Š with the exception of English Language,Literature in English, Art, Music, and subjects taken by private students onlyŠwere grouped into subject clusters (for example, all History papers wereclassified as History), and proportions of children obtaining A grades and Cgrades or better were recalculated for each school type. The school types werethen ranked on the proportion of children obtaining A grades and C grades orbetter in the subject clusters. Rank orders only are presented as the data are notamenable to statistical analysis.RESULTSComparison of academic success in subjects offered by all school typesThe fifteen subjects offered by all school types are listed in Table I. A list of the 47subjects not offered by at least one school category is available from the author.A summary of the chi-square statistics, representing a comparison of theperfomance of the eight school types in each of the fifteen subjects, is provided inTable II. In all but one analysis, Economics grade A passes, the differencesbetween the eight school types are significant. The statistical non-significance ofthe school type differences in Economics grade A passes is due to the combinationof a very low proportion of grade A passes and very few candidates. In eleven ofthe thirty analyses (clearly marked in Table II), the chi-square must be interpretedwith caution. In these eleven analyses, either the proportion of grade A passes oroverall passes, or the number of children writing that subject in one or moreschool categories, or both, was too small to allow meaningful statistical analyses.Rank order of the eight school types by academic successThe ten subjects which were taken by 25,000 candidates or more are EnglishLanguage, Mathematics (Syllabus D), Geography (Central Africa), Shona,46 SCHOOL TYPE & ACADEMIC ACHIEVEMENTTable ISUBJECTS OFFERED BY EVERY SCHOOL TYPE INTHE 1985 'O'-LEVEL EXAMINATIONS,WITH THE TOTAL NUMBER OF CANDIDATESSubject NEnglish Language 101,236Mathematics (Syllabus D) 86,405Geography (Central Africa) 84,788Shona 76,250History (Central and Southern Africa) 62,277Literature in English 50,569Combined Science 41,207General Science (Zimbabwe) 36,142Bible Knowledge (Life of Christ) 31,612Bible Knowledge 25,721Commerce 13,515Principles of Accounts 8,722Human and Social Biology 6,460Biology 5,976Economics 2,349Source: Ministry of EducationHistory (Central and Southern Africa), Literature in English, Combined Science,General Science (Zimbabwe), Bible Knowledge (Life of Christ) and BibleKnowledge. The rank orders of the school types on these subjects are presented inTables Ilia and Nib. The chi-squares calculated using Friedman's analysis ofvariance are 34.41 for grade A passes and 35.56 for overall passes. Both chi-squares have less than a 1 per cent probability of occurring by chance. Table IIIcpresents the overall rank orders of the school types for each of grade A and overallpasses.Comparison of academic success of school types for subject groupsThe rearrangement of the examination subjects into groups produced nine subjectclusters: African Languages, Business Studies, Bible Studies, Geography, History,Mathematics, Other Languages, Practical Subjects and Science. A full break-down of the examination papers included in each cluster is available from theauthor.Tables I Va and IVb present the rank orders of the school types for the ninenew subject clusters for grade A achievement and then overall passes. These dataare presented for heuristic purposes only. They are not amenable to statisticalanalysis because it is possible for one pupil to write, say, two Mathematics papers,rendering the data non-independent to an unknown extent, and also because thefrequency of small numbers underlying the rank orders, would reduce themeaningfulness of the analysis even further.JOSEPHINE JORDAN 47Table IICHI-SQUARE STATISTICS INDICATING THE DIFFERENCES INACADEMIC ACHIEVEMENT OF THE EIGHT SCHOOL CATEGORIESIN THOSE SUBJECTS OFFERED BY EVERY SCHOOL TYPESubject Grade A PassEnglish LanguageLiterature in EnglishBible Knowledge (Life of Christ)Bible KnowledgeHistory (Central and Southern Africa)Geography (Central Africa)EconomicsShonaMathematics (Syllabus D)General Science (Zimbabwe)BiologyHuman and Social BiologyCombined ScienceCommercePrinciples of AccountsDegrees of freedom = 7. Details of the data input and calculations can be obtained from theauthor.* Exercise caution in interpreting these statistics. Chi-sq uares are inapplicable when an expectedfrequency is less than 1 or if 20 per cent of cells (3 or more in this instance) have expected frequenciesof less than 5 (Siegel, 1956, 110)."Only in Economics grade A passes were there no differences between the school categories. Inthis subject, very few pupils obtained A grade passes.DISCUSSIONThe chi-square analyses comparing the achievement of different school types indifferent subjects and at grade A and pass levels, show a school type influence on'O'-level results. The Friedman analyses of variance further show that thedifferences in school achievement are systematic, with Mission schools obtainingthe best results followed by Government Group A schools. District Councilschools obtain the worst results whether A grades or overall passes are considered.The remaining five school types fall between the two extremes.Data presented in Tables IVa and IVb provide a heuristic check of theseresults. Some variations are present when subjects are clustered into groups andthese variations are worth noting, but the post hoc analyses do not materiallychallenge the results of the statistical analysis. (Three notable variations are thatDistrict Council schools fare better when Ndebele is grouped with Shona to formthe cluster African Languages, and when the two major Bible Studies and Sciencepapers are each taken together.)1,554.79193.37*904.69*319.34*874.97551.203.68**774.81831.6929.28*225.97*34.56*182.23*121.8178.234,408.842,474.751,590.961,259.732,570.663,856.7445.24*2,056.132,363.3155.36394.54156.66*906.33*398.60430.4248 SCHOOL TYPE & ACADEMIC ACHIEVEMENTTable WaRANK ORDERS OF SCHOOL TYPES ON ACADEMIC SUCCESS (GRADE A)IN TEN SUBJECTS TAKEN BY 25,000 OR MORE CANDIDATESSubject School Type (k = 8)(10)EnglishLanguageLiteraturein EnglishBibleKnowledgeBibleKnowledge(Life ofChrist)History(CentralandSouthernAfrica)Geography(CentralAfrica)ShonaMathe-matics(Syll. D)GeneralScience(Zimbabwe)CombinedScienceGroupA11212.51.52282.5GroupB54652.556334CommunityCollege2.543.57.5768878DistrictCouncil8854884767Farm/Mine777.57.541.57515.5Mission4212131121Private2.543.535.545642.5Rural/Provincial667.565.573455.5Rj 23.5 43.5 61.5 65 53 18 40 55.5X2 = 34.41 d.f. = 7JOSEPHINE JORDAN49Table MbRANK ORDERS OF SCHOOL TYPES ON ACADEMIC SUCCESS (GRADE COR BETTER) IN TEN SUBJECTS TAKEN BY 25,000 ORMORE CANDIDATESSubject(n=10)School Type (*=8)EnglishLanguageLiteraturein EnglishBibleKnowledgeBibleKnowledge(Life ofChrist)History(CentralandSouthernAfrica)Geography(CentralAfrica)ShonaMathe-matics(Syll. D)GeneralScience(Zimbabwe)CombinedScienceGroupA1122213282GroupB6477565435CommunityCollege2.5635448377DistrictCouncil8856886856Farm/Mine5568634718Mission2.521121Ł21Private7743777643Rural/Provincial4384352564XJ = 35.562452 49.5d.f. = 768 53 14.5 554450SCHOOL TYPE & ACADEMIC ACHIEVEMENTTable IIIcCOMPARISON OF THE SCHOOL TYPES RANKED ONACADEMIC ACHIEVEMENT OVER THE TEN MOST POPULARACADEMIC SUBJECTSSchool typeGrade APassMissionGovernment Group APrivateGovernment Group BFarm and MineRural or Provincial CouncilCommercial CollegesDistrict CouncilTable IVa1234567812756348RANKING OF SCHOOL TYPES BASED ON PERCENTAGES OF A GRADESOBTAINED IN DIFFERENT SUBJECT GROUPSSubject(10)School Type (k = 8)AfricanLanguagesBusinessStudiesBibleStudiesGeographyHistoryMathe-maticsOtherLanguagesPracticalSubjectsScienceGroupA241.52.522435GroupB675.5533Š64CommunityCollege855.565.58188DistrictCouncil484887Š73Farm/Mine7182.54654.56Mission131.5111321Private56345.54.524.52Rural/Provincial327774.5Š17JOSEPHINE JORDAN51Table IVbRANKINGS OF SCHOOL TYPES BASED ON PERCENTAGES OF PASSESOBTAINED IN DIFFERENT SUBJECT GROUPSSubject(n =10)School Type (k = 8)AfricanLanguagesBusinessStudiesBibleStudiesGeographyHistoryMathe-maticsOtherLanguagesPracticalSubjectsScienceGroupA352122454GroupB677654Š65CommunityCollege843443217DistrictCouncil586888Š83Farm/Mine428367546Mission131211321Private764776172Rural/Provincial215535Š38The data presented and analysed in this article replicate and confirm thecontinuing validity of the school type categories, initiated by Irvine and Dorseyand institutionalized in the statistics presented by the University of CambridgeLocal Examinations Syndicate. The consistent rank-ordering of the schoolcategories provides a reliable dimension on which to sample schools in fieldresearch. At the one end of the continuum are Mission schools with consistentlysuperior performance. Contrasted with the Mission category are District Councilschools. Between these two extremes are Government Group A schoolsperforming consistently well, which can be contrasted in turn with GovernmentGroup B schools.The existence of the school type continuum provides unambiguous samplingprescriptions. When it is necessary or desirable to test a theory or procedure in thefull range of Zimbabwean conditions, Mission and District Council schoolsprovide two quasi-experimental conditions which will maximize differences andtherefore provide the most powerful test of the model under investigation. Forpilot studies, or when less rigour is required, it would be possible to contrastsamples derived from Government Group A and Government Group B schools.52 SCHOOL TYPE & ACADEMIC ACHIEVEMENTAcknowledgementsThe Secretary for Education provided permission to conduct the study and thedata were provided by the staff of the Examinations Branch and of the PlanningSection of the Ministry of Education. Their assistance is gratefully acknowledged.ReferencesDORSEY, B. J. 1975 "The African secondary school leaver: Aspirations,academic achievement and post-school employment', in M. W. Murphree(ed.), Education, Race and Employment in Rhodesia (Salisbury, Associationof Round Tables in Central Africa), 13-174.IRVINE, S. H. 1964 'A Psychological Study of Selection Problems at the Endof Primary Schooling in Southern Rhodesia' (London, Univ. of London,Ph.D thesis).IRVINE, S. H. 1969 'Examinations and the economy: An African study', inJ. A. Lauwerys and D. G. Scanlon (eds.), The World Yearbook of Education1969: Examinations (London, Evans), 279-90.SlEGEL, S. 1956 Nonparametric Statistics for the Behavioral Sciences (NewYork, McGraw-Hill).