Die voorspelling van akademiese prestasie van technikon-afstandsonderrigstudente met diverse onderwysagtergronde

Abstract

English: Since the abolition of separate tertiary institutions for different population groups, candidates with diverse academic backgrounds apply for admission at the same tertiary institutions. The increase in student numbers compels tertiary institutions to select applicants. Many South African researchers argue that separate departments of education for the different population groups led to a situation where applicants from the former black secondary schools who received an ineffective school education are being discriminated against by selection procedures based on matriculation results as a predictor of tertiary academic success. Other researchers argue that differences in educational backgrounds are not necessarily detrimental to applicants from disadvantaged educational backgrounds. Cleary's (1968) regression model can incorporate differences in predictor means, criterion means and prediction-criterion correlations for different subgroups. Prediction bias occurs when the criterion performance for a certain demographic group is constantly over- or underpredicted and becomes evident when the regression lines of the subgroups differ. Prediction bias can be removed by computing separate regression lines for different subgroups. Different cut-off points for the different demographic groups involved are then to be determined. The candidates are selected according to their predicted criterion performance. Unbiased predictions are made because candidates with the same predicted criterion performance are either rejected or accepted, irrespective of their demographic group membership. The purpose of the present study was to investigate the validity of matriculation marks as a predictor of the academic performance of first-year technicon distance education students. Secondly, the objective was to determine whether the predictive validity of matriculation results differ for students from advantaged and disadvantaged school backgrounds and finally to investigate the differential prediction of these groups' performance on the basis of matriculation results. The matriculation and first-year results of technicon distance education candidates who enrolled in 1998 at Technicon South Africa in the Free State were used. Matriculation results, high school background and the programme for which the student had registered, were used as predictor variables in the regression equation. The study revealed that the program, for which the student registered, explained 16,7% of the criterion variance. Matriculation results explained 12,5% and secondary school background explained 3,9% of the criterion variance. These results suggest that the programme the first-year student registered for has the greatest effect on his or her tertiary academic performance. Different standards and levels of difficulty between different programmes are most likely the explanation for this finding. The lower than expected percentage of criterion variance explained by matriculation results may possibly be attributed to the longer time interval that exists between school and tertiary education in distance education as opposed to residential education. The lower than expected criterion variance explained by school background can be due to the use of home language as an indicator of high school background. It is possible that some African language speakers indicated English as their home language and could have been categorized incorrectly in the advantaged group. Also, some African language-speaking students could have matriculated from traditionally white matriculation authorities and could have been categorized incorrectly as coming from a non-disadvantaged school background. The correlations between the above-mentioned variables for African (0,0447) and Afrikaans and English speaking candidates (0,1408) were significant on the 1% level. Matriculation performance was thus differentially valid for both groups. The regression equation has different Y-intercepts, but does not significantly differ in slope. No significant interaction between matriculation and group membership was thus found for the groups.

Afrikaans: Met die afskaffing van afsonderlike tersiêre instellings vir verskillende bevolkingsgroepe doen aansoekers met diverse akademiese agtergronde aansoek om keuring by dieselfde tersiêre instellings. Die toename in studentegetalle noodsaak tersiêre instellings om aansoekers te keur. Talle Suid-Afrikaanse navorsers voer aan dat die geskiedenis van afsonderlike onderwysdepartemente vir verskillende bevolkingsgroepe daartoe gelei het dat aansoekers uit die voormalige swart hoërskole 'n benadeelde skoolopleiding ontvang het en daar teen hulle gediskrimineer word in keuringsbesluite wat op matriekprestasie as voorspeller van tersiêr-akademiese sukses steun. Ander navorsers voer weer aan dat die verskille in skoolagtergrond nie noodwendig aansoekers met 'n swakker skoolagtergrond in keuringsbesluite sal benadeel nie. Cleary (1968) se regressiemodel kan verskille ten opsigte van voorspellergemiddeldes. kriteriumgemiddeldes en voorspeller-kriteriurnkorrelasies vir verskillende subgroepe in ag neem. Voorspellingsydigheid kom voor indien 'n voorspeller konstant die kriteriumprestasie van 'n subgroep oor- of ondervoorspel en word statisties sigbaar wanneer die regressielyne vir die twee subgroepe verskil. Voorspellingsydigheid kan statisties verwyder word deur afsonderlike regressievergelykings vir verskillende subgroepe op te stel. Afsonderlike afkappunte wat dieselfde kriteriumprestasie vir die afsonderlike subgroepe aandui, moet bepaal word. Die aansoekers word dan gekeur op grond van hul voorspelde kriteriumprestasie en onsydige voorspellings word verkry deurdat individue met dieselfde voorspelde kriteriumprestasie gekeur of afgekeur word, ongeag hul groeplidmaatskap. Die doel van die ondersoek was om die geldigheid van matrieksimbole by die voorspelling van eerstejaarstudiesukses van technikon-afstandsonderrigstudente te ondersoek. Verder was die oogmerk om na te vors of die voorspellingsgeldigheid van matrieksimbole verskil vir studente uit goeie en benadeelde skoolagtergronde en laastens om die differensiële voorspelling van hierdie groepe se eerstejaarprestasie op grond van matriekprestasie te ondersoek. Die matriek- en eerstejaarprestasie van technikonafstandsonderrigstudente wat in 1998 by T.S.A. in die Vrystaatstreek geregistreer het, is gebruik. Matriekprestasie, hoërskoolagtergrond en die program waarvoor die student geregistreer het, is as voorspellerveranderlikes in die regressievergelyking ingevoer. Uit die ondersoek het geblyk dat 16,7 % kriteriumvariansie deur die program waarvoor hoërskoolagtergrond het 3,9% van die kriteriumvariansie voorspel. Die resultate suggereer dat die program waarvoor die student geregistreer het die grootste rol in die bepaling van eerstejaarsprestasie by studente speel. Verskille in die standaard en moeilikheidsgraad van programme is waarskynlik verantwoordelik hiervoor. Die laer as verwagte persentasie kriteriumvariansie wat deur matriekprestasie verklaar word, kan moontlik toegeskryf word aan die langer tydsperiodes tussen skool- en tersiêre opleiding by afstandsonderrig in vergelyking met residensiële onderrig. Die invloed van skoolopleiding vervaag dus. Hoërskoolagtergrond se laer as verwagte persentasie kriteriumvariansie kan toegeskryf word aan die feit dat huistaal as aanwyser van hoërskoolagtergrond gebruik is. Die moontlikheid bestaan dat sommige Swarttaalsprekende studente hul huistaal as Engels aangedui het en verkeerdelik in die groep met 'n nie-benadeelde skoolagtergrond gegroepeer is. Ook kon sommige van die swarttaalsprekende studente aan tradisioneel blanke matrikulasie-owerhede gematrikuleer het, en dus verkeerdelik in die groep met 'n benadeelde skoolagtergrond gegroepeer gewees het. Die korrelasies tussen matriekpunt en gemiddelde eerstejaarspunt vir die onderskeie skoolagtergronde binne die onderskeie programme was 0,0477 vir studente uit benadeelde en 0,1408 vir studente uit beter skoolagtergrond. Matriekprestasie IS dus differensieël geldig vir die twee groepe. Die regressievergelyking vir die twee groepe toon verskillende Y-afsnitte, maar verskil nie noemenswaardig ten opsigte van die helling vir die twee groepe nie. Daar is dus nie beduidende interaksie tussen matriekprestasie en groeplidmaatskap in die bepaling van technikonprestasie vir hierdie twee groepe nie.