Gender as a Factor in the Prediction of Performance in Botswana General Certificate of Secondary Education Physical Education Examinations by Coursework and Forecast Grades among Senior Secondary School Students

doi:10.4236/ape.2012.21006

Paper Menu >>

Journal Menu >>

Advances in Physical Education

2012. Vol.2, No.1, 32-37

Published Online February 2012 in SciRes (http://www.SciRP.org/journal/ape) http://dx.doi.org/10.4236/ape.2012.21006

Gender as a Factor in the Prediction of Performance in Botswana

General Certificate of Secondary Education Physical Education

Examinations by Coursework and Forecast Grades among

Senior Secondary School Students

Mogomotsi Ramatlala1, Johnson Nenty2

1Department of Curriculum & Instruction, Central Ch in a No rmal University, Wuhan, China

2Department of Educational Foundations, Uni ver sity of Botswana, Gaborone, Botswana

Email: [email protected], [email protected]

Received December 6th, 2011; revised January 10th, 2012; accepted January 20th, 2012

Selection bias is an educational and a social malady which is of concern to every educator especially

educational administrator when allocating educational opportunities to learners based on previous per-

formance. Bias in predicting who passes or who fails and hence in selection exists if the same prediction

equation is used for different groups, for example, for both gender when in fact such prediction, given the

tests involved, is different for male and female learners. As a check for this, there is always the need to

perform differential prediction for the sexes. The purpose for this descriptive study is to investigate gen-

der as a factor in the prediction of performance in BGCSE physical examination by coursework and fore-

cast grades among secondary school students in Botswana. The sample composed of 2292 (1432 males

and 860 females) students who, based on Botswana Examination Council (BEC) records, obtained grades

in coursework, forecast and BGCSE physical education grades for 2005 through 2008. The Pearson cor-

relation and regression analyses done using SPSS revealed that though coursework and forecast grades

significantly predict the BGCSE grades in physical education among senior secondary school students in

Botswana, based on each year’s data, gender does not influence such prediction significantly, but for cu-

mulative data across all the years it does.

Keywords: Differential Prediction; Gender Bias in Selection; BGCSE Physical Education Examination;

Coursework Grades; Forecast Grades; Botswana

Introduction and Background

Education is the single most expensive enterprise globally and

especial ly so in Bo tswa na whe re go ver nment spen ds ab out 26 % of

her annual budget on this enterprise. It is common practice to de-

termine academic readiness and hence perform selection of learners

for the next level of education based on the results of their per-

formance in previous related examinations. Conducting examina-

tions within and at the end of ea ch and every school year is a part

of the school curriculum and in Botswana, like in other countries;

learners take public examinations to determine their academic

standing at each level of education. Coursework grades from con-

tinuous assessment and Mock examination or forecast grades are

often used to predict students’ readiness to take the final school

public examination. Public examination grades are commonly used

as a measure of acade mic achievement for a given level of e duca-

tion. Achievement at a given level of education is often used to

determine or predict the probable level of achievement i n the next

level of education or at the workplace.

In Botswana at the end of basic education, and at the end of

secondary education, students are expected to sit for public ex-

aminations such as the Junior Certificate in Education (JCE) and

Botswana General Certificate of Secondary Education (BGCSE)

examinations respectively. BGCSE examination started in 1993

in partnership with the University of Cambridge Local Examina-

tion Syndicate (UCLES). Later when it was realized that the

country was in sufficient control of the education standards and

could develop and administer examinations across all levels,

localization of marking examinations took place as a recommen-

dation by the Report of the National Commission on Education

(RNPE) of 1993. BGCSE physical education examinations were

first administere d in 2005 taking the form of pract ical and theory

based examinations.

Theoretical Foundation

The mean of scores on several measures of the same ability

is likely to give a more valid indication of that ability than the

result of a one-shot measure of the ability. It is on this premise

that continuous assessment is advocated in schools as it pro-

vides for opportunity for improvement based on occasional

feedback. But if the instruments used in the repeated measure-

ments are differentially invalid for the different groups involved,

then the sum of the scores would not provide an unbiased

measure of the ability under consideration. The case of con-

tinuous assessment of students contributing to their final grade

in the graduating examination is a well argued one (Little,

1992). Teachers are in stronger position to assess the attainment

of their students than the external examiner who has only the

evidence of a few hours work to draw from. Little highlighted

that the rationale for coursework in GCSE has been that it can

explore aspects of physical education knowledge and practice

M. RAMATLALA ET AL.

which short time written examinations questions cannot reach

and do not provide enough time to investigate in any depth.

In any setup, academic achievement is generally assumed to

be an indication of the student ability to succeed in the educa-

tional program. The first difficulty with any consideration of

these purposes is that the functions and values of public ex-

aminations and coursework are not constant. Both are deter-

mined by the wider purpose of education and the changing

needs of the society involved (Robbins, 1997). Assuming the

purpose of coursework, given that it is properly administered

and weighted, is determined by the overt function of the ex-

amination which might not perfectly reflect the needs of the

society. In that case, the realization of these purposes is unlike-

ly to be achieved (Ramatlala, 2009). In other words to what

extent could one count on what the student has done in school,

as reflected by his/her performance, in admitting him/ her to the

next level of education, or in employing him/her in an area that

requires the same or rela t ed skills?

Iramaneerat (2007) stated that different countries use differ-

ent measures to evaluate academic readiness of students. Bot-

swana’s Ministry of Education in collaboration with Botswana

Examination Council recommended class or theory based ses-

sions in addition to practice-based sessions for performance of

coursework in physical education. According to Ramatlala

(2009) teachers assess the students in practical activities and

grade them for assessment of learning purpose. The assessment

is carried at each and every teaching and practical activity. The

marks from physical activities are combined with theory based

assessment at the end of two years of high school to determine

the learning outcome.

According to Robbins (1997), examination is a process of

measuring, scrutinizing and investigating the knowledge, un-

derstanding and capability of the candidate. Coursework is any

sort of performance or work produced to meet the purpose of

the curriculum. Forecasting, on the other hand, is to estimate or

calculate in advance, the likely performance level of the learn-

ers, and could be determined through analysis of relevant data.

The expectation is that students getting good grades in course-

work and forecast test should be in a position to attain good

grades during final examinations. Coursework and forecast

grades represent academic achievement over a period of time

and given that they are based on the same content with BGCSE,

they can be appropriately used as predictors of BGCSE final

examination grades.

Studies on coursework have revealed a significant predictive

strength of coursework in final examination (Masole & Utl-

wang, 2005; Thobega & Masole, 2008). Daniels and Schouten

as cited by Adeyemi (2010) argued that a prediction of a future

examination result could be made with reasonable success on

the basis of the result of an earlier examination and that grades

may serve as prediction measures and as criterion measures.

This boils down to finding the validity of the measures used

to determine the likely achievement of the learners in the crite-

rion measure. According to Young (2001), validity is not the

inherent characteristics of a test but it must be determined for

each use of a test and for the population of interest. Mock and

forecast grades are used in Botswana to predict who would earn

what level of pass during Botswana General Certificate of Sec-

ondary Education Examination (BGCSE) (Masole & Utlwang,

2005; Thobega & Masole, 2008). Depending on the predictor

and criterion variables involved, sometimes the results from

such analysis are often used to select students for advancement

to the next level of education. But some extraneous variables

especially gender, school location, type of school, etc. may

pose extraneous influences that bring about invalidity in such

exercise. It might lead to differential prediction which results in

bias in the selection purpose. This depends on the predictive

validity of the variables used in such exercise.

Bias in Testing and in the Use of Test Results

Bias in testing exists when the influences of extraneous sour ce s

in a testing situation result in members of one group performing

significantly higher or lower than members with the same abi-

lity but from other groups taking the same test. In other words,

when examinees with different group membership have been

determined to have the same ability in what the test was de-

signed to measure, are matched up, members from one group

systematically influenced by the extraneous variable show sig-

nificantly different performance in the test items. So, extrane-

ous sources that influence performance on a test can make the

test biased against one group or the other because the test items,

influenced by extraneous sources, function differentially across

members of groups taking the test. In other words, scores that

results from a biased test have different meaning for different

groups who took the same test (Nenty, 1979).

The use of test results can also be biased if they differentially

over-predict a criterion of interest for one group while under-

predicting it for another group of examinees. That is, if the

same test (or te sts) has si gnifican tly diffe rent pre dictive val idity

for the different groups involved in the testing. Therefore, pre-

dictive bias has to do with differential validity which leads to

differential prediction. This study is concerned with the second

meaning of bias. Selection bias is an educational and a social

malady which is of concern to every educator especially educa-

tional administrator when allocating educational opportunities

to learners based on previous performance. According to Kyei-

Blankson (2005), questions about differential prediction, there-

fore, are questions related to whether the prediction models

obtained for the different groups of examinees are different.

Such questions are generally approached by comparing regres-

sion systems for equality of regression slopes and equality of

regression intercepts in the respective prediction equations

across groups of interest (p. 48).

Statement of the Problem

Every examination results serve the fundamental purpose of pre-

dicting the readiness of the learner for the next level of education or

suitability for employment. Some examination might lack predic-

tive validity and hence predict such suitability falsely and some-

times differently for different groups of students, for example

males and females. T he predictive validity of certain examinati ons

has been a matter of concern to many researchers (Adeyemi, 2008).

Prediction bias is a source of and might lead to social injustice

which a selection process might unknowingly commit. Studies

have shown that certain examinations such as Scholastic Aptitude

Test (SAT) validly predicts university grades, while certain low-

quality examinations could not effectively predict performance at

higher level examinations (Kobrin et al., 2008; Young, 2001). This

is more so for teacher made classroom tests and some public ex-

aminations in Africa. Using grades from school coursework ex-

aminations and forecast grades from mock examinations to predict

who passe s or who f ails and he nce for sel ection is l ikely to lea d to

biased decision-making especially if the same prediction equation

M. RAMATLALA ET AL.

is used for both gender whereas such prediction might be different

for male and female learners. This is unfair for the group whose

performance is under pred icted an d unfairly limits their opportunity

to be selected or allocated educational opportunities based on the

results of p rev ious exam ina tions .

Purpose and Objectives

This study seeks to determine whether coursework and fore-

cast grades significantly predict performance in BGCSE phy sical

education examinations and equally so for male and female stu-

dents. Some earlier studies (Thobe ga & Masole, 2008; Masole &

Utlwang, 2005) have revealed course work and forecast grades to

be good predictors of BGCSE grades and in agriculture grades,

but none have been found trying to determine gender-based pre-

dictive validities of these measures across many years and for

physical education grades. Hence this stu dy aims at investigating

the predictive strength of these t w o variables on BGCSE phy sical

education examinations.

Given the fear of gender-based differential prediction by these

two variables the st udy spec ificall y aims at determining the extent

to which gender influence s the pre diction o f students pe rforma nce

in BGCSE physical education examination by both coursework

and forecast grade. H ence the objectives of the study ar e to:

 determine the extent to which performance in school-based

coursework and forecast grades predict performance in

BGCSE physical education examinations for all, male and

for female students;

 determine the extent to which gender influences the predic-

tion of BGCSE physical education performance by course-

work and forecast grades.

Research Hypothesis

In the null form, these state that:

Ho1: Coursework and forecast grades do not significantly

predict BGCSE physical education performance for male and

female students.

Ho2: Gender has no significant influence on the ability of

coursework and forecast grades to predict BGCSE physical

education performance.

Methodology

The population for the study was students who took BGCSE

physical education examinations f or the four years: 2005 throu gh

2008. There were a total of 3190 of such students. Out of this

number, those that came from schools with grades in coursework,

forecast grades as well a s in BGCSE were 2292 (1432 males a nd

860 females) students. These served as the sample for the de-

scriptive study.

Measures

Secondary data were used for the study, and were retrieved

with permission from Botswana Examination Council (BEC)

academic records. In BEC, intensive panel-based content analy-

sis and face validation is carried by each subject panel. So the

grades were deemed valid for use in the study.

Procedure

Data were coded and entered into the computer and analysis

were done by carrying out Pearson correlation coefficient and

three multiple regression analyses. All the analyses were done

using SPSS version 16 for Windows. The prediction model for

coursework and forecast grades were determined by fitting the

values of the relevant parameters in the linear regression model.

The predictor variables in the regression analysis were the stu-

dents’ coursework and forecast grades and BGCSE as the crite-

rion variable. The significance level for testing the hypotheses

set at .05 for all statistical tests.

Data Analysis and Interpretation of Results

A descriptive analysis of the research data gave the means

and standard deviation as well as the Pearson correlation values

among the three variables of the study for all students as well as

for males and females separately (se e Table 1).

To test the first hypothesis, a multiple regression analysis

was done with BGCSE grades as the dependent variable and

coursework grades and forecast grades as the independent va-

riables (see Table 2). The result showed that coursework grades

accounted for 42% of the variability of grades in the BGCSE

while with forecast grades it accounted for about 51% of such

variance. The analysis gave an F-value of 44.668 which given 2

& 2289 degrees of freedom was found to be significant be-

yond .01 alpha level. Hence among secondary school students

in Botswana the null hypothesis was rejected and the research

hypothesis that both coursework grades and forecast grades

significantly predict the final BGCSE grades in physical educa-

tion was retained. The analysis gave a prediction equation as

indicated in Formula 1.

To test the second hypothesis, the same analysis was done

for data collected for males and females students separately

(see Tables 3 and 4). With a predictive validity of .473 (R2

= .473) for males and .551 for females, the two variables were

found, in each case, to account for a substantial variability in

the BGCSE grades respectively. The regression formula for

both gender are given in Formula 2 for males and 3 for females.

The resulting predictive validities were then compared using

Fisher’s transformed Z-test (see Table 5) to determine whether

ender has significant influence on the prediction of BGCSE g

Table 1.

Mean, standard deviati o n and inter-correlation matrix of variables in the study.

All Subjec t s (n = 2292) Female (n = 860) and Male ( n = 1432)1

Variable Mean (SD) BGCSE Forecast

Grade Course-

work Mean (SD)BGCSE Forecast

Grade Course-

work Mean

(SD)

BGCSE 4.72 (1.03) 1.000 .421* .648* 4.49 (1.01)1.00 .399* .619* 4.86 (1.03)

Forecast Grade 4.81 (1.41) .421* 1.000 .185* 4.77 (1.38).467* 1.00 .166* 4.84 (1.43)

Coursework 4.32 (1.18 .648* .185* 1.000 4.00 (1.24).663* .212* 1.00 4.52 (1.09)

*p< .01; 1Mea ns and correl ation values for males a re above the diagonal a nd that for females are bel ow the diagonal.

M. RAMATLALA ET AL.

Table 2.

Prediction of BGCSE performance in physical education using coursework and forecast grades for all students (n = 2292).

Change Statistics

Model R R

Square Adjusted R

Square Std. Error of

the EstimateR Square Change F Change df1 df2 Sig. F Change

1 .648a .421 .420 .7912 .421 1661.863 1 2290 .000

2 .717b .515 .514 .7243 .094 44.668 1 2289 .000

Sum of Squares df Mean Square F Sig.

Regression 1273.121 2 636.561 1213.388 .000b

Residual 1200.842 2289 .525

Total 2473.963 2291

Unstandardized Coefficients Standardized Coefficients

Constant/Variables B Std. Error Beta

t-Value Sig.

(Constant) 1.360 .071 19.072 .000

Coursework F i nal Grades (CW)

Forecast Grade (FG) .522

.229 .013

.011 .591

.312 39.881

21.063 .000

.000

Depende nt Variable: Student BG C S E final grades.

aPredictors: (Constant), coursework final grades; bPredictors: (C onstant), coursework final grades, forecast grade s. BGCSE = 1.360 + .522 CW + .229FG.

Table 3.

Prediction of BGCSE performance in physical education using coursework and forecast grades for all male students (n = 1432).

Change Sta t i s tics

Model R R

Square Adjusted R

Square Std. Error of

the Estimate R Square Change F Change df1 df2 Sig. F Change

1 .619a .383 .383 .8072 .383 888.3785 1 1430 .000

2 .688b .473 .473 .7460 .090 244.952 1 1429 .000

Sum of Squares df Mean Square F Sig.

Regression 715.118 2 357.559 642.442 .000b

Residual 795.328 1429 .557

Total 1510.446 1431

Unstandardized Coefficients Standardized Coefficients

Constant/Variables B Std. Error Beta t-Value Sig.

Constant 1.380 .100 13.801 .000

2 Coursework Final Grades (CW) .484 .018 .568 29.204 .000

Forecast Grade (FG) .256 .014 .305 15651 .000

Depende nt Variable: Student BG C S E final grades.

aPredictors: (Constant), coursework final grades; bPredictors: (C onstant), coursework final grades, forecast grade s. BGCSE = 1.380 + .537 CW + .219FG.

Table 4.

Prediction of BGCSE performance in physical education using coursework and forecast grades for all female students (n = 860).

Change Statistics

Model R R

Square Adjusted R

Square Std. E rror of the

Estimate R Square ChangeF Change df1 df2 Sig. F Change

1 .663a .439 .439 .7613 .439 672.196 1 858 .000

2 .742b .551 .550 .6819 .111 212.338 1 857 .000

Sum of Squares df Mean Square F Sig.

Regression 488.315 2 244.157 525.053 .000b

Residual 98.518 857 .465

Total 886.833 859

Unstandardized Coefficients Standardized Coefficients

Constant/Variables B Std. Error Beta t-Value Sig.

Constant 1.360 .102 13.292 .000

2 Coursework Final Grades (CW) .483 .019 .590 25.192 .000

Forecast Grade (FG) .251 .017 .341 14.572 .000

Dependent Variable: Student BGCS E final grades.

aPredictors: (Constant), coursework final grades; bPredictors: (C onstant), coursework final grades, forecast grade s. BGCSE = 1.360 + .483 CW + .251FG.

M. RAMATLALA ET AL.

Table 5.

Validity of coursework and forecast grades in predicting BGCSE in physical education from 2005 to 2008.

From Simple Regression From Multiple RegressionUnique Contribution of Forecast Grades

Coursework Grade Forecast Grade

Year Gender N

R Z-Value 2

R Z-Value Coursework &

Forecast GradesZ-Value 2

R -

RZ-Value

Male 366 .407 .190 .472 .068 –1.18

2005 Female 272 .442 –1.04 .271 –1.37 .528 –1.00 .086 –.97

Total 638 .457 .209 .521 .064 –1.61

Male 317 .512 .357 .555 .043 –.78

2006 Female 213 .544 –.50 .354 .03 .592 –.66 .048 –.78

Total 530 .562 .368 .601 .039 –1.01

Male 403 .453 .277 .560 .107 –2.18*

2007 Female 205 .454 –.02 .245 .05 .583 –.41 .129 –1.89*

Total 608 .459 .266 .574 .115 –2.89*

Male 346 .566 .170 .659 .093 –2.00*

2008 Female 170 .627 –1.03 .230 –.09 .745 –1.82 .118 –1.89

Total 516 .591 .192 .691 .100 –2.89*

Male 1432 .383 .159 .473 .090 –3.11*

Total Female 860 .439 –1.71 .218 –1.91 .551 –2.53* .112 –3.25*

Total2 2292 .421 .177 .515 .094 –4.39*

*p < .05; Critical Z = 1.96.

grades using coursework grades and forecast grades. This gave a

Z-value of –2.53 (p = .005) which, in absolute value, is higher

than the critical Z-value of 1.96 (α = .05). Hence the second null

hypothesis was rejected for the combined data, meaning that the

prediction lines for BGCSE from coursework and forecast grades

for males and females have significantly different slopes.

A similar comparison for each of the four years separately

did not result in any significantly different predictive validity

for male and female students (see Table 5). Therefore, for the

combined data for all the four years, a significant gender-based

differential prediction was observed, but for each of the years,

gender was not found to have a significant influence on the

prediction of BGCSE grades using coursework grades and

forecast grades among secondary school students in Botswana.

A closer study of the related regression equations shows that

based on data for the four years combined coursework grades

and forecast grades significantly under-predict the BGCSE

grades in physical education for males and over-predict for

females students. Such significant trend was not found when

the parameters for regression equations for each of the four

years were examined (see Table 5 ).

In summary therefore coursework and forecast grades sig-

nificantly predict the BGCSE grades in physical education

among secondary school students in Botswana and gender is

found to influence such prediction significantly across the years

but not for each of the years. A summary of the effect sizes and

prediction differential prediction indicators are presented in

Table 5.

For each of, and for all the years 2005 to 2008, both course-

work grades and forecast grades individually are significant

predictors of BGCSE grades in physical education (see Table 5).

But coursework grades are always significantly superior to

forecast grades in predicting BGCSE grades in this subject.

Generally such prediction is always significantly improved if

both independent variables are used together. For physical edu-

cation, whether coursework grades or forecast grades are used

singly or combinely to predict BGCSE grades for each of the

years, gender does not constitute a significant biasing factor in

such prediction, but predictive validity for females is always

higher than that for males (see Table 5). A significant sign of

such effect appears when the prediction is based on a combined

four years’ data, but it is rare to use more than one year’s data

for such exercise.

Discussion of Findings and Recommendations

Fair selection has a lot of implications for social justice, access

and for equality in the identification and development of potential

through education. Selection based on incorrect prediction may

deprive some learners of the oppo rtunity to continue their educa-

tion or channel them into areas to which they are not naturally

endowed. This may result in inefficient utilization of manpower

as well as of resources. Invalid prediction often results in un-

der-predicting the criterion variable for one group of students

while over-predicting it for the other(s). This often leads to bias

in selection, which is of very high social and educational concern.

Given the cumulative data for the four years under considera-

tion, in this study the two measures used coursework and fore-

cast grades—significantly predict performance at BGCSE

grades in physical education and based on data cumulated over

four years, such prediction differs between male and female

secondary school students in Botswana. This confirms the dif-

ferential prediction often found in similar studies (Young, 2001).

The very high percentage (about 40%) of criterion variance

accounted for by continuous assessment grades for each of the

three (male, female, & total) predictions is indicative of the pre-

dictive validity of this class exercises given BGCSE in physical

education. This tells a lot about the quality of continuous as-

sessment in physical education in Botswana schools. The nature

of this assessment is its strength. It takes into consideration all

M. RAMATLALA ET AL.

that BGCSE examination involves, plus incorporating the prac-

tical work aspects of the subject. Over and above the contribu-

tion of continuous assessment, forecast grades accounted for

some significant percentage (9% to 11%; see Table 5) of the

criterion variance in each of the three predictions based on the

four years’ data. This is partly due to the significant amount of

variance it shares with continuous assessment (2.75% to 4.50%)

in the three predictions and partly due to it seemingly lack of

content validity as it does not involve the practical aspects of

physical education. But for these, given its close resemblance to

the criterion, one would have expected forecast grades from

Mock examination to account for more of the criterion variance.

Predicting BGCSE performance in physical education from

grades in coursework and forecasting in Botswana differenti-

ates between male and female students and hence lead to bias in

any selection if such prediction is based on cumulative data.

The significantly different prediction validities (.473 for males

and .551 for females; Z = –2.53, p < .05) is indicative of bias in

prediction if and when cumulative data across the years are

used.

The story was not quite the same if the prediction is based on

yearly data. In none of the y ears is such differentially significant

prediction observed (see Table 5). For each of these years,

coursework grades and forecast grades were seen to individually

and combinely predict grades in BGCSE physical education sig-

nificantly with predictive validities ran ging from .457 to .591 fo r

coursework grades, .192 to .368 for forecast grades, and

from .521 to .691 for both predic tors combined. Though the sig-

nificant predictive validities of each and both coursework and

forecast grades persisted across each of the four years, the sig-

nificant differential prediction was not observed for any of the

years considered separately. One can then say that the observed

significant differential prediction for the combined four years’

data is an artifact of the increased sample size. In other words, the

likelihood of improved predictive validity was enhanced by com-

bining data for four years and hence increasing the sample size.

It could be concluded that in Botswana senior secondary

schools, for students’ performance in physical education be-

tween the years 2005 to 2008, coursework grades and forecast

grades, individually and combinely, were significant predictors

of BGCSE performance. They also showed gender-based sig-

nificant differential prediction, over-predicting the criterion for

females while under-predicting for male students. But when the

analyses were done for each of the four years, such gender-

based significant differential prediction was not observed for

any of the four years.

Recommendations

Given that the results of predicting BGCSE grades may be

used in flagging students whose BGCSE grades might be re-

viewed (Masole & Utlwang, 2005), Botswana Examination

Council should ensure that the prediction of grades in this ex-

amination should be done with measures that will ensure no

gender-based differential prediction. This would mean taking

measures and procedures that enhance the predictive validities

of each of the predictor variables through improving their con-

tent validities at the classroom levels.

Similar studies as to the validity of coursework, forecast grades

and even other measures in predicting BGCSE grad es in different

subjects should be carried out. Other variable-selecting methods

of multiple regression analy sis could be used instead of the step-

wise method used in this st udy. Such studies should also be car-

ried out with other variables like ethnicity, location and socio-

economic levels as factors in the prediction of performance in

BGCSE physical education examination by coursework and

forecast grades among secondary school students in Botswana.

REFERENCES

Adeyemi, T. O. (2010). Credit in mathematics in senior secondary

certificate examinations as a predictor in educational management in

universities in Ondo and Ekiti States. Nigeria. Middle-East Journal

of Scientific Research, 5, 235-244.

Adeyemi, T. O. (2008). Predicting students’ performance in senior

secondary certificate examinations from performance in junior sec-

ondary certificate examinations in Ondo State, Nigeria. Humanity &

Social Sciences Journal, 3, 26-36.

Akbari, R., & Alivar, N. K. (2010). L2 teacher characteristics as pre-

dictors of students’ academic achievement. The Electronic Journal

for English as a Second Language, 13, 1-22.

Gardner, D., & Deadrick, D. L. (2008). Understanding of performance

for US minorities’ cognitive ability measures. Equal Opportunity In-

ternational, 27, 455-464. doi:10.1108/02610150810882305

Iramaneerat, C. (2007). Rater errors in a clinical skills assessment of

medical students. Evaluation & the Health Professions, 30, 266-283.

doi:10.1177/0163278707304040

Kobrin, J. F., Patterso n, B. F ., Shaw, E. J ., Mattern, K. D., & Barb uti, S.

M. (2008). Validity of the SAT® for predicting first-year college

grade point average. College Board Report No. 2008-5.

Kyei-Blankson, L. S. (2005). Predictive validity, differential validity,

and differential prediction of the subtests of the medical college ad-

mission test. Unpublished Doctoral Dissertation, Athens, OH: Ohio

University.

Little, C. (1992). GCSE coursework: A retrospective assessment.

Teaching Mathematics and Its Applications, 11, 56-63.

doi:10.1093/teamat/11.2.56

Masole, T. M., & Utlwang, A. (2005). The reliability of forecast grades

in predicting students’ performance in the final Botswana General

Certificate of Secondary Education Examinations.

Nenty, H. J. (1979). An empirical assessment of the culture fairness of

the Cattel Culture Fair Intelligence Test using the Rasch Latent Trait

Measurement Model. Unpublished Doctoral Dissertation, Kent, OH:

Kent State University.

Ramatlala, M. S. (2009). The validity of coursework scores in predict-

ing performance in Botswana General Certificate of Secondary

Education Physical Education examinations among senior secondary

school students in Botswana. Unpublished Master’s Thesis, Gabo-

rone: University of Botswana.

Robins, K. (1997). What in the world’s going on? In P. D. Gay (Ed.),

Production of culture/cultures of production. London: Sage/Open

University.

Thobega, M., & Masole, T. M. (2008). Relationship between forecast

grades and component scores of the Botswana General Certificate of

Secondary Education Agriculture.

Young, J. W. (2001). Differential validity, differential prediction, and

college admission testing: A comprehensive review and analysis.

College Board Research Report No. 2001-6. New York: College En-

trance Examination Board.

Young, J. W. (1991). Gender bias in predicting college academic per-

formance: A new approach using item response theory. Journal of

Educational Measurement, 28, 37-47.

doi:10.1111/j.1745-3984.1991.tb00342.x