Research Predicting Factors That Affect the Student’s Academic Performance

PREDICTING FACTORS THAT

AFFECT STUDENTS’
ACADEMIC PERFORMANCE
BY USING DATA MINING
TECHNIQUES
Najmus Saher Shah
College of Computer Science & Information Systems
Institute of Business Management, Karachi
Abstract

The study attempts to investigate the influence of factors on

students’ academic performance by comparing the accuracy of
different classifiers. The result will be useful in identifying good
as well as weak students who may perform poorly and will be
potential dropouts. Students are categorized in five groups
according to their performance: “Very Good”, who have a high
probability of succeeding; “Good” students, who are above
average and with little efforts, may succeed with good grades;
“Satisfactory” students, who may succeed; “Below satisfactory”
students, who require more efforts to succeed; and “Fail”, who
have a high probability of dropping out. The data set comprised
of 231 students out of 637 from a leading degree institute of
Karachi: registered in the graduate programs of year 2009. Weka,
an open source data mining software is used to explore the
influence of factors on predicting students’ academic
performance. Dataset is applied to different classifiers of Weka :
J48,RandomForest, RepTree and BFTree of Decision Tree, Bayes
and NaiveBayes of BayesNetwroks, Logistic and RBFNetwork
functions and JRip rule. The study also shows that re-sampling
of data was a critical step which is the reason of the success of
the study.

631 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

Keywords: Academic performance, data mining, Decision Tree,

Bayesin Netwroks.
JEL Classification: Z000

I. Introduction

To identify potential drop outs of the institute’s graduate

program is a complex process mostly due to the fact that students
coming from different backgrounds have certain characteristics
as well as perceptions and apprehensions of the world of the
university and their actions embody them.

Students’ failure to integrate and acquire good grades

are considered to be one of the main factors but many researchers
have also suggested that there are various other factors that may
affect students progress at the university level.

Predicting successful and unsuccessful students at an

early stage of the degree program help academia not only to
concentrate more on the bright students but also to apply more
efforts in developing programs for the weaker ones in order to
improve their progress while attempting to avoid student
dropouts.

II. Literature Review:

Abdul Mannan(2007) explores the relation between

academic and social integration and also the impact of student’s
integration in academics and social activities on their academic
performance. A sample of 2400 full time undergraduate students
enrolled in 3rd term of 2002 at the University of Papua New Guinea
was considered. Social integration was studied on the basis of i)
informal social contact with academic staff ii) extracurricular
activities and iii) peer group interaction. Academic integration
consisted of two criteria: i) academic staff concern for students’

PAKISTAN BUSINESS REVIEW JANUARY 2012 632

Research Predicting Factors That Affect the Student’s Academic Performance

development and learning and ii) students informal contact with

academic staff on academic matters. Data analysis was done by
using the SPSS package. A strong negative relationship was
found between academic and social integration which indicates
that less integration in the social domain of the university was
compensated by higher academic integration leading to students’
persistence in continuing their studies. It was found that
students’ persistence differs according to their subject area of
studies.
Ali (2010) conducted a study based on 324 students in
different cities of Pakistan and concluded that student motivation
is an element that determines students’ attitude towards the
learning process. Data was collected with the help of a
questionnaire. Intrinsic motivation and extrinsic motivation i.e.
motivation derived from external rewards were selected as
independent variables and academic performance was selected
as the dependent variable. Ali (2010) found that academic
performance is positively influenced by intrinsic motivation and
negatively influenced by extrinsic motivation. The aspiration
within and various external factors help in motivating students
to work hard. Highly motivated students’ success in academics
is stronger as compared to others.

Nattavudh, A.Vignoles(2009) showed that there is a

significant gap in the dropout rate between students who have
a strong family background and those who have a weak
background. Various factors like income, occupation, education,
social status can determine a background. It was found in the
study that Students from higher socio-economic and more
educated backgrounds have lower rates of dropouts.

Similarly, Ishitani (2003) has shown that first generation

students (whose parents did not graduate from college) tend to
dropout more as compared to others. He investigated the impact
of independent variables such as race, gender, family income

633 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

and high school GPA on the behavior of a dropout student. He

also explored whether their influence vary at different points of a
student’s academic career. Five types of outcomes were
considered; continue stopout, dropout, transfer, and graduate.
Exogenous (race, gender, high school GPA, family income etc)
and time dependent variables (college GPA, financial aid, and
athletic status) are assumed to affect an individual student’s
outcome in each time period. A sample of college students in fall
of 1995 with enrollment status for 5 academic years was collected.
Information of student characteristics used is based on a freshman
survey conducted during the 1995 freshman orientation. An event
history model (survival analysis) was used which is a powerful
statistical tool used to study events and their causes (Blossfeld,
2002). Survival function is estimated by product-limit-estimation
and exponential models. Product line estimation model considered
three groups of students with different parental educational
backgrounds. The result showed that students whose parents
both of whom were college graduate achieved the highest survival
rates. Group of students who had one college-educated parent
had a slightly lower rate as compared to those whose both of
whose parents are college graduate in 1st and 2nd semester but this
gap widened through 3rd to 6th semesters. The exponential model
assumes that the effects of explanatory variables are constant
and change proportionally over time (Ishitani, 2003). The
exponential model ‘s result showed that the coefficient estimate
for first-generation student was 0.253 indicating that first-
generation students had attrition rates higher than those who
had two college–educated parents. Thus, first generation students
had 29% higher rate of departure than the reference group.

Another survival analysis modeled was the piecewise

exponential model. Piecewise exponential model with period-
specific effects model assumes that the effects of explanatory
variables are constant within each period but vary across different
periods (Ishitani, 2003). The risk of departure for first generation

PAKISTAN BUSINESS REVIEW JANUARY 2012 634

Research Predicting Factors That Affect the Student’s Academic Performance

students were 71% higher. Minority students have lower attrition

rate rather than their counterparts in first and second years.
Students belonging to lower income had 49%higher risk of
leaving in the first year but decrease to 26% higher as compared
to high income group students in the second year. Overall
outcomes in the study were consistent with the previous studies.
First –generation students were more likely to depart than were
their peers and risk of departure among first generation students
varied over time.

Data mining techniques relate to the field of database

and can be useful in predicting or forecasting various hidden
trends in the data. It is an emerging field originally applied to the
commercial transactions. It is being used in addition to traditional
statistical methods in higher education for finding causes for
students’ behavior, their decision of transferring or quitting and
identifying students at risk of dropping out. Data mining
techniques can be very helpful in determining students’
performance.

Data mining is comprehensively defined by the Gartner

group(2000) as” “the process of discovering meaningful new
correlations, patterns, and trends by sifting through large
amounts of data stored in repositories and by using pattern
recognition technologies as well as statistical and mathematical
techniques”

Rubenking (2001) view data mining as “the process of

automatically extracting useful information and relationships from
immense quantities of data. In its purest form, data mining doesn’t
involve looking for specific information. Rather than starting
from a question or a hypothesis, data mining simply finds patterns
that are already present in the data.”

635 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

Data mining techniques like decision tree, neural

network, Bayesian network were used in the studies in predicting
performance to come up with results that can more accurately
predict students’ performance as compared to traditional statistical
methods.

Decision trees are a collection of nodes, branches, and

leaves. Each node represents an attribute; this node then split
into branches and leaves until the data are classified to meet a
stopping condition.

Bayesian network is based on decision theory. It is a

branch of probability and statistics which investigates how to
minimize risk and loss when making decisions based on uncertain
information. It is a graphical model that encodes relationships
among variables that it models.(Rahel Bekele et al(2005).

Neural network is a set of interconnected nodes. Nodes

at one layer are connected to all nodes at the next layer. The layer
which receives input is called input layer and the final layer is
called output layer. In between these two layers are hidden layers.
The variables of the problem are acted upon and weighted by the
entry layer, which then transmits this information to the hidden
layers; these combine all the information into hidden layers; these
combine all the information into a single value which is passed to
the exit node and which acts as a kind of estimated value for the
decision variable.( Vandamme et al ,2007)

Logistic regression is a type of predictive model based

on linear regression technique that associates a conditional
probability score with each data instance. (Richard J. Roiger)

Gerben W. Dekker (2009) used data sets of students’

past education as well as university grades and related data
collected from the electrical engineering department of Eindhoven

PAKISTAN BUSINESS REVIEW JANUARY 2012 636

Research Predicting Factors That Affect the Student’s Academic Performance

University of Technology to test their impact on students’

performance and determine whether they can help in predicting
performance. The study was aimed to differentiate successful
and unsuccessful students as well as those who are at risk of
dropping out as early as possible in the bachelors program.
Various simple and sophisticated data techniques were used
and their results were compared. The overall result shows that
simple classifier gives result with accuracy between 75 to 80%.

Thai Nghe et al (2007) explored the use of data mining

techniques in predicting students’ performance. He used the
decision tree approach and Bayesian network and also compared
the accuracy of two data mining techniques algorithms applied
on the students of two different academic institutes. The Asian
Institute of Technology (AIT) data sets included students’
records and GPA at the end of the second year to predict the
students’ rate of performance in the third year. The other set of
Can Tho University (CTU) in Vietnam included students
admission data to predict GPA at the end of the first year. Using
the data mining tool i.e Weka, it was found that predictions for
the CTU data set were more accurate as compared to AIT due to
larger number of records for CTU. Decision tree algorithm
consistently performed better than the Bayesian network
algorithm .Other findings were that accuracy in smaller classes
was much lower than in larger classes. This was overcome by
using resample function in Weka. It significantly improved the
accuracy of prediction where the original data had a much smaller
sample size.

Some students take more years than usual to complete

their degree program. Universities have to anticipate this type of
behavior in students and have to come up with policies and
rules to discourage this type of behavior but before this they
have to know the reasons or factors influencing students
conduct.

637 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

To estimate student retention and degree-completion

time, Herzog (2006) used data mining techniques which included
decision tree, neural networks and logistic regression. He
compared the prediction accuracy of these methods by generating
data from three different sources. i)institutional student
information for student demographics, academic, residential and
financial aid information, ii) American college test’s students
profile section for parent income data, National student clearing
house for identifying transfer-year enrollment of new full time
freshmen who started in fall semester of 2000 to 2003. Multitopolgy
neural network; creates several networks in parallel based on a
specified number of hidden layers and nodes in each layer. It
performed significantly better in identifying dropout but perform
poorly in estimating who is likely to transfer.

Neural network with hidden layers achieved 25%

improvement in predicting when analysis included only new
students. This is also true for the decision tree and pruned neural
network models. Data mining algorithms worked better with a
large set of exploratory predictors used to estimate degree
completion time. Where time to graduate is estimated for new and
transfer student simultaneously, pruned neural network and C5.
Decision tree performed better.

Pruned neural network starts with a large network of

layers and nodes as specified and removes (prunes) weakest
nodes in input and hidden layers during training.

C5.0 uses the 5.0 algorithm to generate a decision tree or

rule set based on the predictor that provides maximum information
gain. The split process continues until the sample is exhausted.
Lowest –level splits are removed if they fail to contribute to model
significance.

PAKISTAN BUSINESS REVIEW JANUARY 2012 638

Research Predicting Factors That Affect the Student’s Academic Performance

Mario Jadric et al (2010) attempted to find the dropout

rate among students. Transaction data on students was collected
through the faculty of economics Information system within the
autonomous subsystem student service. Data mining was
conducted in SAS 9.1 Enterprise Miner (software) by using
techniques of decision tree, neural network, and logistic
regression. The model developed was based on SEMMA
(sampling, exploring, modifying, modeling and assessment)
methodology. It was found that women dropout comparatively
less than men and students with better entrance ranking dropout
less. The analysis separated the causes of students’ dropout
and it also determined the typical profile of the student inclined
to dropout.

William R. Vetch’s (2004) study was conducted to

identify the relationship between independent variables such as
academic performance, age, gender, ethnic group; socio-
economic status, grade point average on the dropout behavior
in high schools. He used a decision tree model to test his
hypotheses. He utilized extant data resources and all variables
were extracted from district electronic databases. High school
students recorded as “dropped” (no transfer record) over the
course of 2001-2002 academic year were matched with a random
sample of non- dropouts. The study revealed that academic
performance is the most related variable to dropping out
behaviors. Older students are more likely to dropout as compared
to younger students. The tree exhibits a certain ability to predict
which students may drop out of school.

Vandamme et al (2007) uses three sets of factors i)history

of student (his identity, socio-family past, academic past, age,
gender), ii)Student’s involvement in studies (participation in
optional activities, meeting with lecturers and iii) Student’s
perception(views on academic context, professors, course to
determine their influence on the students performances in

639 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

academics. The study used students’ data survey to collect data.

The questionnaire (comprised of 42 questions) was distributed
to first year students and based on it a database was created in
which each student is described according to attributes
(explanatory variables X) .Each student is also assigned with a
risk-of –failure category(high, medium, low risk of failures)and so
created dependent variable Y. Several data mining techniques
like decision tree (based on Shannon’s entropy and ID3
algorithms), Neural network (multi layer), Linear Discriminant
Analysis were used and SAS /Enterprise Miner software was
used in applying these techniques. 20% variables showed
significant correlations with academic success and 80% rate of
correct classification in predicting the success or failure of
students.

The objective is to not only to identify the factors that

have an effect on the students’ academic performance but to
compare the accuracy of different classifiers in predicting
students’ performance. The goal is to categorize students’
performance in five groups : “Very good”, with a high probability
of succeeding; “Good” students, who are above average and
with a little more effort can succeed with good grades ;
“Satisfactory” students, who may succeed; “Below Satisfactory”
students, who require more efforts to succeed; and “Fail” , who
have a high probability of dropping out. Thai Nghe et al (2007) in
his study also grouped students in different classes according to
their performance and compared and evaluated the accuracy of
Decision Tree and Bayes algorithms for predicting performances.

III. Data:

The attributes selected are identified and tested by

different studies reviewed above (section II). Factors that are
part of the study are given in the Appendix Table 1along with the
variables that determine these factors.

PAKISTAN BUSINESS REVIEW JANUARY 2012 640

Research Predicting Factors That Affect the Student’s Academic Performance

Students registered in Fall 2009 belonging to BBA,

BS programs of a private business university in Karachi
are targetted. For the study, the collection of data has
been done through online questionnaire .Complete
questionnaire of 231 students out of 637 were received.
Data including list of students enrolled and current GPA
i.e. GPA of first year was collected from the Institute.

Online questionnaire was developed (Appendix

3)which is based on the findings of the literature review
.The Likert scale rating 5 steps has been used for questions
21-84. 84 items are included to ascertain the affect of these
variables on students’ GPA. Some of the questions used
in the questionnaire are part of other studies also
(Appendix 4&5).Items representing a particular factor are
explained in (Appendix 2)

Weka is selected as a data mining tool as not only

is it an open source software but also because it is used
by many researchers. Weka (Waikato Environment for
Knowledge Analysis) is a popular suite of machine learning
software written in Java, developed at the University of
Waikato, New Zealand. Weka, is a free software available
under the GNU General Public License and it supports
several standard data mining tasks, more specifically, data
preprocessing, clustering, classification, regression,
visualization, and feature selection.

Students’ GPA data collected from the Institute

was in continuous form which has to be transformed in to
d i s c r e t e c a t e g o r i e s . T h e r e f o r e , s t u d e n t s ’ GPA w a s
categorized in five groups as shown in the Table 1.

641 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

Table 1: Categories of GPA

GPA GROUP
3.78-4 Very Good
3.33-3.77 Good
2.67-3.32 Satisfactory
1.5-2.66 Below Satisfactory
0-1.49 Fail

After categorizing the data, I checked for missing data,

fortunately there was none. First generation student was one of
the variables of factor: however family background was not
considered in data analysis as majority of the students have
misunderstood the question resulting in incorrect answers. Since
dataset consists of 87 items, therefore, data was further analyzed
by selecting relevant items defining a particular factor by using
Weka’s feature of selecting attributes. Weka is comprised of many
attribute selection selectors. InfoGain Attribute Evaluator was
selected to distinguish relevant attributes by ranking them
according to their significance in determining the dependent
variable as well as to reduce the size of prediction.

The information gained, with respect to a set of examples

is the expected reduction in entropy that results from splitting a
set of examples using the values of that attribute. This measure is
used for identifying those attributes that have the greatest
influence on classification. Thai Nghe et al (2007)

Before the data is subjected to attribute selector,

unsupervised discretization on numerical attributes is applied as
some of the classifiers like tree and rule learners work well with
discretized attributes.

Best five variables which have the highest significance

in determining a particular factor were selected and the rest
discarded. This practice was applied for all factors. All together 8

PAKISTAN BUSINESS REVIEW JANUARY 2012 642

Research Predicting Factors That Affect the Student’s Academic Performance

factors were used in the research for determining their significance

in predicting students’ performance.

Weka provides various algorithms grouped in different

classifying methods. On the basis of studies reviewed, most
commonly used and good predictor algorithm of Weka classifiers
(with their default settings) were selected as shown below in
Table 2. The aim is to compare these algorithms in predicting
students’ performance.

Table 2 : Data mining techniques used in the study

Decision Trees: J48(C4.5),RandomForest,BFTree, RepTree

Functions: Logistic, RBFNetwork
Rule: JRip
Bayesian Network: BayesNet, NaiveBayes

J48 decision tree and Random Forest are used in the

study by Vandamme et al (2007). Thai Nghe et al (2007), Al-
Radaideh et al (2006), and Othman et al (2007) used J48 decision
tree and BayesNetwork classifiers. Affendy et al (2010) also used
j48, BayesNetwork along with Naïve BayesNetwork, RepTree,
BFTree, RBF Network and Logistic function. Gerben
W.Dekker(2009) used rule learning JRip along with J48 decision
tree and Random Forest, Bayes Network along with Naïve Bayes
Network, and Logistic function.

All the classifiers are run on the dataset using 10-fold

cross validation for estimating generalization performance. It was
used to evaluate the prediction accuracy.

Cross-validation divides the data in to fixed number of

folders. For example, if a data is divided into n folders f1,f2,…,fn,
then cross validation method asks f1 to train(f2,f3,…, fn). In second

643 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

stage f2 train (f3, f4, …, fn) and so on until the folders are trained to
prepare the certainty.

The classification accuracy of each classifier is

summarized in Table 3 to compare their relative performance.

Table 3: Simulation result of each algorithm

Original Data Resample Data

Kappa Kappa
Correctly Incorrectly Correctly Incorrectly
Predicted Statistics Statistics
Algorithm Classified Classified Classified Classified
GPA (Total Instances Instances Instances Instances
Classes Instances,231) %(value) %(value) %(value) %(value)
15.368
49.3506 50.6494 84.632 (71)
J48(C4.5) (114) (117) 0.0467 (391) 0.7429
47.619 52.381 92.4242 7.5758
5 classes:(Very Good, Good,Satisfactory,Below

(110) (121) (427) (35)

RandomForest 0.0734 0.8742
47.619 52.381
(110) (121) 79.4372 20.5628
BFTree 0.0203 (367) (95) 0.6514
50.2165 74.4589 25.5411
Satisfactory,Fail)

49.7835 (116) (344) (118)

RepTree (115) 0.0623 0.5633
46.3203 53.6797 74.8918 25.1082
(107) (124) (346) (116)
JRip 0 0.5704
45.8874 54.1126 66.0173 33.9827
(106) (125) (305) (157)
RBFNetwork 0.0667 0.4215
38.0952 61.9048 66.4502 33.5498
(88) (143) (307) (155)
Logistic 0.0405 0.4443
51.0823 48.9177 58.658 41.342
(118) (113) (271) (191)
BayesNet 0.1307 0.2857
50.6494 58.0087 41.9913
49.3506 (117) (268) (194)
NaïveBayes (114) 0.0879 0.2658

PAKISTAN BUSINESS REVIEW JANUARY 2012 644

Research Predicting Factors That Affect the Student’s Academic Performance

IV. Results:

From the obtained results, it was found that the

classification accuracy for all the classification algorithms is not
high. This indicated that the data set was small and not sufficient
to generate a classification model of high quality in terms of
accuracy. On inspection of confusion matrices from tables 4-
12(original data column), it was found that accuracy for smaller
classes was much smaller as compared to larger classes in every
classifier output. For example, in Table 4: classifier J48 decision
tree, there was no prediction of the number of students failing
but 84% students have been predicted as satisfactory
performers. Even the class of “Very Good” was not considered
in confusion matrices of every classifier, the reason being that
those students who participated in the survey, their first year
CGPA did not fall in this group. Therefore, confusion matrices
for all classifiers only represented four classes.

To resolve the problem of imbalances in the dataset,

data was resampled by using Weka resample function. This
function oversamples the minority class and under samples the
majority class in order to create a more balanced distribution for
training the algorithms. By looking at the table, it is clearly
observed that predictions using the re-sampled data set are
significantly more accurate for all classifiers.

On comparing classifiers, Table 3 (Resmapled Data

column), RandomForest turned out to be the most effective
classifier in predicting students’ performances with accuracy
92% followed by J48 decision tree 84%, BFTree 79%, RepTree
74% and JRip rule 74%. The least effective is the BayesNet and
NaïveBayes with 59%and 58% accuracy respectively. Even
though before resampling, BayesNet had the highest accuracy
rate of 51% in predicting students’ performance and and
NaiveBayes was same as J48 tree 49%. This shows that

645 PAKISTAN BUSINESS REVIEW JANUARY 2012

 Table 4: J48 Tree

Original Data Re-Sampled Data

Actual Class Below Below
Satisfactory Satisfactory Good Fail Satisfactory Satisfactory Good Fail

Satisfactory 97 16 2 0 194 22 1 1
Below Satisfactory 65 16 0 0 26 156 3 0
Good 23 3 1 0 7 7 30 2
Fail 5 3 0 0 1 1 0 11
%Hit 84% 20% 4% 0% 89% 84% 65% 85%

Table 5: RandomForest
Original Data Re-Sampled Data

PAKISTAN BUSINESS REVIEW JANUARY 2012

Actual Class Below Below
Satisfactory Good Fail Satisfactory Good Fail
Satisfactory Satisfactory
Predicting Factors That Affect the Student’s Academic Performance

Satisfactory 76 36 3 0 203 13 1 1
Below Satisfactory 46 33 2 0 8 176 1 0
Good 18 8 1 0 3 6 37 0
much as it affects the others algorithms on the dataset.

Fail 3 4 1 0 0 1 1 11
%Hit 66% 41% 4% 0% 93% 95% 80% 85%
Research

resampling does not affect these algorithms’ ability to predict as

646
Research Predicting Factors That Affect the Student’s Academic Performance

RandomForest’s correctly predicted students in

classes “Good” and “Fail” as 80% and 85% (Table 5)
respectively on resampled data. It showed that resampling
of data have a significant impact on predicting capability
of a classifier. Before that, the classifier predicted poorly
for these classes. Even the predictions for “Satisfactory”
and “Below Satisfactory” classes have also improved
considerably.

RepTree correct prediction for “Satisfactory”

class (Table 6) decreases by 1% on resampled data as
compared to the prediction on original data but its overall
performance is much better.

The accuracy in the predictions of BFTree, JRIP

rule and RBF function, performances increases as shown
in tables 7,8 and 9. Logistics Function (Table 9) prediction
performance increases more for “Good” and “Fail” classes
as compared to “Satisfactory” and “Below Satisfactory”
classes.

647 PAKISTAN BUSINESS REVIEW JANUARY 2012

 Table 6: RepTree

Original Data Re-Sampled Data

Actual Class Below Below
Satisfactory Good Fail Satisfactory Good Fail
Satisfactory Satisfactory
Satisfactory 95 18 2 0 179 33 5 1
Below Satisfactory 61 19 1 0 41 142 2 0
Good 22 4 1 0 14 13 19 0

Internal Factors
Fail 7 1 0 0 5 2 2 4
%Hit 83% 23% 4% 0% 82% 77% 41% 31%

Table 7: BF Tree

Original Data Re-Sampled Data

Actual Class Below Below

PAKISTAN BUSINESS REVIEW JANUARY 2012

Satisfactory Satisfactory Good Fail Satisfactory Satisfactory Good Fail

Satisfactory 90 25 0 0 184 28 5 1
Predicting Factors That Affect the Student’s Academic Performance

Below Satisfactory 61 20 0 0 34 150 1 0

Good 24 3 0 0 10 5 31 0
Fail 5 3 0 0 6 5 0 2
%Hit 78% 25% 0% 0% 84% 81% 67% 15%
Research

648
 649
Table 8: JRip
Research

Original Data Re-Sampled Data

Actual Class Below Below
Satisfactory Satisfactory Good Fail Satisfactory Satisfactory Good Fail

Satisfactory 89 24 1 1 178 36 1 3
Below Satisfactory 62 18 1 0 42 141 2 0
Good 25 2 0 0 20 3 23 0
Fail 6 2 0 0 8 0 1 4
%Hit 77% 22% 0% 0% 82% 76% 50% 31%

Table 9: Logistic Function

Original Data Re-Sampled Data

Actual Class Below Below
Satisfactory Satisfactory Good Fail Satisfactory Satisfactory Good Fail

Satisfactory 46 42 16 11 151 57 7 3
Below Satisfactory 26 39 8 8 71 112 1 1
Good 17 6 3 1 6 7 33 0
Fail 5 2 1 0 0 1 1 11
%Hit 40% 48% 7% 0% 69% 60% 72% 85%

PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance
 Table 10: RBFNetwork

Original Data Re-Sampled Data

Actual Class Below Below
Satisfactory Satisfactory Good Fail Satisfactory Satisfactory Good Fail

Satisfactory 74 30 10 1 159 53 4 2
Below Satisfactory 46 30 5 0 60 121 4 0
Good 17 7 2 1 13 13 20 0
Fail 5 2 1 0 5 2 1 5
%Hit 64% 37% 7% 0% 73% 65% 43% 45%

Table 12: BayesNet

Original Data Re-Sampled Data

Actual Class Below Below

PAKISTAN BUSINESS REVIEW JANUARY 2012

Satisfactory Satisfactory Good Fail Satisfactory Satisfactory Good Fail

Satisfactory 80 32 2 1 152 61 2 3
Predicting Factors That Affect the Student’s Academic Performance

Below Satisfactory 42 37 2 0 72 106 6 1

Good 18 8 1 0 28 9 9 0
Fail 6 2 0 0 4 3 2 4
%Hit 70% 46% 7% 0% 70% 57% 20% 31%
Research

650
Research Predicting Factors That Affect the Student’s Academic Performance

Kappa Statistics which is used to compare different

classifiers predicting performance is quiet low for every classifier
while predicting on the original dataset (Table 3). In fact, for JRip,
it was zero. But after resampling, Kappa Statistics for all classifiers
have improved significantly. RandomForest has the highest 0.8742
followed by J48 decision tree 0.7429. Bayesian networks have the
lowest, again, confirming the fact that resampling does not affect
the predicting capabilities of Bayesian networks on the data set.

Kappa statistics is used to assess the accuracy of any

particular measuring cases, it is usual to distinguish between the
reliability of the data collected and their validity.

Factors that are strong and significant predictors of the

students’ performances are academic integration, family
background and social integration.

Academic integration i.e. students’ involvement in their

studies by spending more hours on study and its related issues,
students contentment with their choice of study program as well
as the pressure of workload that they can cope with easily, make
their interaction with studies more constructive.

Family background with mother’s education, father’s

income and occupation along with the size of family ensure
students’ contentment at home, which reflects in their
achievements.

At the academia, interaction with faculty and peers,

involvement in sports and other extracurricular activities help
students in their performances. In fact, students’ motivation within
one self as well as friendly surroundings boosts their inspirations
to succeed. This all, support students to work hard and achieve
their goals.

651 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

Individual characteristics and extrinsic motivation have

the least effect on student’s performance which shows that these
are not the significant predictor.

IV Conclusion:

This study shows that datamining techniques applied

can be used by universities and higher education in determining
best students so that resources available can be affectively be
utilized in helping and guiding these students to achieve success,
especially in selecting students for scholarships and other means
of financial assistance. This will also help these academic bodies
and board of studies in developing meaningful programs that
motivates and encourage those students who got the potential
to excel but need assistance to progress. Academia can also
develop programs that build up close relationship between
teachers and the students as shown in the study that interaction
between these two factors help the latter one immensely in their
success.

In case of identifying best students (good) and those

(Satisfactory), who with little help can fall in good students’
class, highest prediction accuracy is that of Random Forest at
80% and 93% respectively. Over all, Decision tree classifiers
predict these students more accurately as compared to other
classifiers. To identify those students who are not performing
well and those who are at the risk of failing (below satisfactory)
and eventually be droppedout (fail), again RandomForest’s
accuracy is the highest with accuracy of 95% and 85%
respectively. Looking at the performance of all the classifiers on
the dataset of students, it is evident that Decision tree classifiers
are better, in terms of accuracy, in predicting students’ academic
performance.

PAKISTAN BUSINESS REVIEW JANUARY 2012 652

Research Predicting Factors That Affect the Student’s Academic Performance

Acknowledgments:

I thank Rizwan Ahmed (Senior Lecturer, Institute of

Business Management) for his time, valuable input and assistance
in doing this research.

References:

Afzal, H.,Imran,A., Khan,M.A. and Kashif,H. (2010) A study of

university students’ motivation and its relationship with their
academic performance, International Journal of Business and
Management,, Vol. 5, 4.

Affendey, L.S., Paris,I.H.M., Mustapha,N., Sulaiman, Md.N., and

Muda,Z.(2010) Ranking of Influencing Factors in Predicting
Students’ Academic Performance.Information Technology Journal
Vol. 9, 4.

Al-Radaideh, Q. A., Al-Shawakfa,E.M. and Al-Najjar, M.I(2006)

Mining Student Using Decision Trees, The International Arab
Conference on Information Technology.

Bekele, R. and Menzel,W.(2005) A Bayesian Approach to Predict

Performance of A Student(BAPPS): A Case with Ethiopian
Students.Artificial Intelligence and Applications: IASTED
International Conference Proceedings.

Bellaachia, A.,Guven, E.(2005). Predicting Breast Cancer

Survivability Using Data Mining Techniques. Department of
Computer Science The George Washington University.

Bouckaert, Remco R.(2004) Bayesian Network Classifiers in Weka,

1st September

653 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

Dekkar,W.G., Pechenizky,M. and Vleeshouwers, M.J.(2009)

Predicting Students Drop Out: A case study. Submitted to the
2nd Int. Conf. on Educational Data Mining (EDM ’09).

Hall, M. and Holmes,G.(2002) Benchmarking Attribute Selection

Techniques for Discrete Class Data Mining. Working paper
Series, 18 April.

Herzog, S. (2006) Estimating student retention and degree-

completion time: Decision trees and neural networks vis—vis
regression. New Directions for Institutional Research,
Vol.2006,Issue 131

Ishitani, T.T. 2003. A Longitudinal Approach To Assessing

Attrition Behavior Among First-Generation Students: Time-
Varying Effects Of Pre-College Characteristics.Research In Higher
Education. 2003, Vol. 44, No.4,pp.433-449.

Jadric, M., Garaca,Z. and Cukusic,M.(2010) Student Dropout

Analysis With Application Of Data Mining
Methods.Management,Vol.15, pp.31-46

Karegowda, A.G., Manjunaath,A.S. and Jayaram, M.A.(2010)

Comparative study of attribute selection using Gain Ratio and
Correlation Base Feature Selection, Internationl Journal of
Information Technology And Knowledge Management, July-
Dec,Vol. 2, pp. 271-277.

Kirkby,R., Frank, E. and Reutemann, P. (2007)WEKA Explorer

User Guide for Version 3-5-5,26 January.

Kokkodis,M. and Akabay,M. CS 235 Project Report: UCSD Data

Minning ContestMannan, MD.A.( 2007) “Student Attrition And
Academic And Social Integration: Application of Tinto’s Model at
the University of Papua New Guinea.”, Higher Education, Vol 53,
pp. 147-165
PAKISTAN BUSINESS REVIEW JANUARY 2012 654
Research Predicting Factors That Affect the Student’s Academic Performance

Nge, Thai.( 2007) “A Comparative Analysis Of Techniques For

Predicting Academic Performance.” In Proceedings of 37th conf.
on ASEE/IEE Frontiers in Education.

Ogor, E.N.(2007) Student academic performance monitoring and

evaluation using datamining techniques. Electronics, Robotics and
Automotive Mechanics Conference, CERMA 2007, pp. 354–359

Othman,M.F. and Moh Shan Yau, T.( 2007) Comparison of Different

Classification Techniques Using WEKA for Breast Cancer. IFMBE
in university drop out.

Rubenking, N. “Hidden Messages.” PC Magazine, May 22, 2001,

20(10), 86–88

Sahay, A., Mehta, K. and QMS,LLP(2010) Assisting Higher

Education in Assessing, Predicting, and Managing Issues Related
to Student success: A web-based software using data ,mining
and quality function deployment, Academic and

Business Research Institute Conference, las Vegas Vandamme,

J.P. and Superby,J.F(2007) “Predicting Academic Performance by
Data Mining Methods.”Education Economics, Vol.15 No(4),
pp.405–419.

Veitch, W.R. (2004) “Identifying characteristics of high school

dropouts: Data mining with a decision tree model”, Presented at
the Annual Meeting of the American Educational research
Association held on April at San Diego, CA.

Kappa at http://www.dmi.columbia.edu/homepages/chuangi/
Kappa

Gartner Group. “The GartnerGroup CRM Glossary.” [http://

www.gartnerweb.com/public/static/hotc/hc00086148.html].

655 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

Appendix Table 1: Factors of the Study

Factors that are part of the model are given in the table below
along with the variables that determine these factors.

Factors Variables of the study Variables used in other studies

Family Father’s education Nattavudh, A.Vignoles(2009), Ishitani
background (2003), Vandamme et al (2007), Mario
Jadric et al (2010), Gerben
W.Dekker(2009)
Mother’s education Nattavudh, A.Vignoles(2009), Ishitani
(2003), Mario Jadric et al (2010),
Vandamme et al (2007), Gerben
W.Dekker(2009)
Father’s occupation Nattavudh, A.Vignoles(2009), Thai Nghe
et al (2007), Vandamme et al (2007),
Ishitani (2003), Mario Jadric et al (2010),
William R. Veitch (2004)
Family income Nattavudh, A.Vignoles(2009), Ishitani
(2003), Herzog (2006)
First generation Ishitani (2003), Mario Jadric et al (2010)
Parental martial stability Vandamme et al (2007)
Family size Vandamme et al (2007)
Availability of computer and net at home None as added as a new variable to test
social Quality and quantity of student’s relationships Abdul Mannan(2007)
integration with peers
Interaction with faculty Abdul Mannan(2007)
Extracurricular involvements Abdul Mannan(2007)
Academic GPA Abdul Mannan(2007), William R. Veitch
Integration (2004)
Hours spend on academic or extra-curricular Vandamme et al (2007)
activities
Entrance test score Thai Nghe et al (2007), Mario Jadric et al
(2010)
Enrolled in preferred course Thai Nghe et al (2007), Mario Jadric et al
(2010), Gerben W.Dekker(2009),
Herzog(2006)
Field of study Thai Nghe et al (2007), Mario Jadric et al
(2010), Gerben W.Dekker(2009),
Herzog(2006)
Missed classes Vandamme et al (2007)
Individual Gender Ishitani (2003), Thai Nghe et al (2007),
Characteristics William R. Veitch (2004), Vandamme et al
(2007), Mario Jadric et al (2010),
Herzog(2006), Nattavudh,
A.Vignoles(2009),
age Vandamme et al(2007),Gerben
W.Dekker(2009),Herzog(2006),Thai Nghe
et al (2007),William R. Veitch (2004)
Career Educational expectations or career Herzog(2006)
expectations expectations
and Goal
Commitment
Satisfaction Student’s satisfaction with the college Herzog(2006)
with experience i.e Resources, Facilities
Institutional (computer,net,library) and Structural
Characteristics arrangement
Individual Intrinsic Imran Ali et al, 2010
Motivation Self-exploration, Altruism
Extrinsic
Rejection of Alternative Options, Career and
Qualifications, Social Enjoyment, Social
Pressure

PAKISTAN BUSINESS REVIEW JANUARY 2012 656

Research Predicting Factors That Affect the Student’s Academic Performance

Appendix 2:

Theory support that relates the questionnaire to questionnaires of reviewed

studies.
Age

Question 3 is framed to find out the age of a student. Impact of age on student’s
performance was tested by Vandamme et al(2007),Gerben
W.Dekker(2009),Herzog(2006),Thai Nghe et al (2007),William R. Veitch (2004).

Gender

Question 4 is asked to find whether student’s persistence to continue education differ

with gender. This factor was included in the studies of Nattavudh, A.Vignoles(2009),
William R. Veitch (2004), Terry T. Ishitani (2003), Vandamme et al(2007), Thai Nghe et al
(2007), Herzog(2006)

Family back ground

This dimension is added as it has been investigated in many studies. Questions 10-16
are related to this dimension.

Parents’ education

Question 5and 6 are included to gather information about the qualification of father
and mother to see whether their education has an impact on students’ performance.
Terry T. Ishitani (2003), Gerben W.Dekker(2009), Vandamme et al(2007), Mario Jadric et
al (2010) and Nattavudh, A.Vignoles(2009) all included questions in their questionnaire
related to parents education.
First generation

Question 7 is included because Terry T. Ishitani (2003)and Mario Jadric et al (2010)

used this measure to find its impact on students’ persistence to perform well.

657 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

Father’s occupation and Family Income

Question 8 and 9 are included as Nattavudh, A.Vignoles(2009)

included this aspect along with the annual income of the family
in his studies. Annual income of the family is used as a measure
by Terry T. Ishitani (2003), Thai Nghe et al (2007), Vandamme et
al (2007)Mario Jadric et al (2010) and William R. Veitch (2004) in
their studies.

Parental marital status

Question 10 is asked to see the impact of parents’ relation on

students’ performance. It was investigated by Vandamme et al
(2007) in his study.

Family size

Question 11 is asked to explore this factor influence on student’s

performance. Vandamme et al (2007) used these measures in their
studies.

Academic Integration

This dimension is included in almost all the studies reviewed.

Field of study and preferred program

Question 12 and 13 are included to find out the impact of choice of

study and enrollment in an unwanted study program on a student’s
performance. Gerben W.Dekker(2009),Herzog(2006),Thai Nghe et al
(2007), Mario Jadric et al (2010) included this factor in their studies.

Missed classes and hours spend on academic activites outside

college

PAKISTAN BUSINESS REVIEW JANUARY 2012 658

Research Predicting Factors That Affect the Student’s Academic Performance

Question 14 is added to find out how much a student try hard to acquire knowledge.
Vandamme et al(2007) in their studies used this factor to investigate its level of impact
on performance.

Question 15 is added to explore the difference in student’s performance who spend

less or more time on academic activities.

Computer and Internet

Since computer and net has become so much part of a student’s life , therefore, Question
16and 17 are added to see the impact of facilities of computer and net on the academic
progress of a student.

Work load

Questions 46-49 are included to see the impact of workload of assignments and quizzes
on student’s performance.

Social Integration

This dimension is generally used as a measure to predict students’ performance.

Interaction with faculty

Questions 21-30 are asked to find the importance and influence of teaching method as
well as teacher student relation on students’ performance. These questions have been
adopted by MD.Abdul mannan(2007) in his questionnaire.

Extracurricular activities

Questions 31-35 are framed to find out how well a student is integrated in his social
life. These questions were adopted by MD.Abdul mannan(2007) and Mike taffe in
their questionnaires . Influence of extracurricular activities on students’ performance
was also tested by Vandamme et al(2007) and Terry T. Ishitani (2003).

659 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

Peer group interaction

Questions 36 -40 are included to find whether students relation and interaction with
his peers has an impact or not on his academic performance. Abdul mannan(2007) did
include questions related to this measure in his questionnaire.

Goal commitment and career expectations

Questions 41-43 are framed to find whether better career prospects motivates student
to work hard in completing their degree. Herzog (2006) included this factor to estimate
student retention.
Question 44 and 45 are formulated to find whether student’s determination let a student
to persist and aspire for degree completion.

Student satisfaction with his selected institute

Questions 50-64 are included to test whether students satisfaction with the facilities
and services provided by the institute and institute’s atmosphere have an impact on
students performance. This factor was again included in the study carried out by
Herzog (2006) in estimating student retention.

Motivation

This dimension is included to explore the influence of student’s own motivation on his
performance.
Questions 65- 84 will investigate the impact of this factor. Imran Ali et al(2010) in their
study tried to find out to what degree this measure has an influence on students
academic progress.

Current GPA

This factor is included and will be collected from the institute. Gerben W.Dekker(2009),
Thai Nghe et al (2007),William R. Veitch (2004) and Herzog(2006) included this factor to
find student retention and dropout rates.

PAKISTAN BUSINESS REVIEW JANUARY 2012 660

Research Predicting Factors That Affect the Student’s Academic Performance

Appendix 3: QUESSTIONNAIRE

Predicting factors that affect the student’s academic performance by using data
mining techniques
1. Name : _________________
2. Student ID: _________________

Individual Characteristics:
3. Age: a. Below 18 b. Between 18 and 20 c. Between
21 and 25

4. Gender: a. Male b. Female

Family Background:
5. Father’s level of education
a. Middle school b. Intermediate c. Graduate d.
Masters e. Doctorate
6. Mother’s level of education
a. Middle school b. Intermediate c. Graduate d.
Masters e. Doctorate
7. Are you the first one in your family to attend bachelors level institute a.
YES b. NO
8. Father’s occupation
a. Business b. services c. Government officer d.
others
9. Annual Family Income
a. <2,40,000 b. 2,40,000- 4,80,000 c. 4,80,00 0-
7,20,000 d. 7,20,000-9,60,000 e. Above 9,60,000
10. Parental marital status
a. Married b. Divorce c. Widower
11. How many members in your family?
a. 3 b. 4 c. 5 d. 6 e. Others_________

Academic Integration:
12. Field of study:
a. Computer science b. Finance
c. Marketing d. Economics
e. Human Resource Management f. Information Technology

661 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

13. Are you enrolled in your preference program a. YES b. NO

14. How often do u miss your classes? a. Daily b. Once a week c. Twice a
week
15. Hours spend on academic activities
a. Less than 2 hr b. Between 2 to 4 hrs c. More
than 4 hrs
16. Do you have access to computer at home? a. YES b. NO
17. Do you have access to internet at home? a. YES b. NO

Social Pressure (Motivation)

18. Your family expects you to graduate? a. Yes. b. No
19. Do you try to live of others expectations a. Yes. b. No
20. Is it expected of you to enroll for an advance degree when, or if, you
complete your graduate degree?
a. Yes. b. No

Social Integration:
Select the most appropriate one (Strongly
disagree, Disagree, Neutral ,Agree, Strongly Agree)

Interaction with faculty

21. Faculty are genuinely concerned in my academic work
22. Faculty are genuinely interested in teaching
23. Faculty are interested in alleviate my academic weakness
24. Faculty are always available for obtaining information
25. Faculty are accessible to discuss matters of intellect
26. Faculty are accessible to discuss career goals
27. Faculty feedback make you work harder
28. Faculty have positive influence on personal growth
29. Faculty promote good relationship
30. I have regular contact with teachers outside of class

Extracurricular activities
31. I participate in clubs and organizations
32. I participate in sports and cultural events
33. I attended a meeting of a club, organization
34. Interpersonal relationship enhance personal growth
35. Interpersonal relationship expand intellectual growth

PAKISTAN BUSINESS REVIEW JANUARY 2012 662

Research Predicting Factors That Affect the Student’s Academic Performance

Peer group interactions

36. Students help in personal problems
37. I get on well with other students
38. I sat around in the student center talking with other students
39. I prefer to study with other students i.e. in group
40. Studies bodies and groups promote friendship

Career Expectations And Goal Commitment:

41. The job prospects for the major are promising
42. The major has well paid jobs
43. There is value of the university education I am receiving
44. I give up easily on difficult projects/assignments.
45. I have goals in life that I try to achieve

Academic Integration:
Work load
46. The workload on my degree courses is manageable and do not apply
unnecessary pressure.
47. Degree courses do not apply unnecessary pressure on me as a student.
48. The volume of work on my courses means I can always complete it to my
satisfaction.
49. I am generally given enough time to understand the things I have to learn

Students’ Satisfaction:

Campus:
50. Transport facilities provided are sufficient
51. Facilities (stationary, photocopier, printing) provided by tuck-shop is
sufficient.
52. Friendly atmosphere and pleasant learning environment
53. Clean and nice campus grounds
54. Class room size and number of students in a class is satisfactory.
55. Class environment as a whole interest you in studying

Library:
56. The library is a good, quiet place for studying and the study rooms are
great for working on group assignments.
57. Vast array of books in library, and usually have available what I’m looking
for

663 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

Computer and net facilities:

58. Number of computers available in the library is sufficient.
59. Internet facility is good and is always available at the institute

Staff:
60. Friendly and approachable staff

Security:
61. You feel secure and safe at the campus.

Institutional commitment
62. I feel a sense of pride about my campus.
63. I am able to experience intellectual growth here.
64. There is a commitment to academic excellence on this campus.

Motivation

I attend university…
65. because I don’t know what else to do.
66. to understand myself better.
67. to gain valuable skills for my career.
68. because its fun place to be.
69. because I genuinely want to help others.
70. because it’s a better alternative than working.
71. because I want to explore new ideas.
72. because I enjoy the social life.
73. because other people have told me I should.
74. because I want to contribute to society.
75. to avoid being unemployed.
76. because I want to challenge myself.
77. because it gives me something to do.
78. because it will help set up my future career.
79. because of the social opportunities.
80. because I want to improve the world situation.
81. because I don’t have any better options.
82. because I love learning.
83. so I can get a better job.
84. because its a great place to develop friendships.

PAKISTAN BUSINESS REVIEW JANUARY 2012 664

Research Predicting Factors That Affect the Student’s Academic Performance

Appendix 4: Similar Questions used in survey of study by

Mannan,A.(2007)

(A) Academic staff concern for students development and teaching

1. Attended departmental meetings

2. Genuinely concerned in my acad. work
3. Willing to spend time outside class
4. Genuinely interested in teaching
5. Interested in alleviate my academic weakness

(B) Informal contact with academic staff on academic matters

6. Always available for obtaining information

7. Accessible to discuss matters of Intellect
8. Positive influence on personal growth
9. Accessible to discuss career goals
10. I am satisfied with opportunities

(C) Informal social contact with academic staff

11. Accessible to discuss campus etc issue

12. Interested for socialization
13. Accessible to solve personal problems
14. Involve promoting good relationship
(D) Extracurricular activities

15. Participation in clubs and organizations

16. Participate in sports and cultural events
17. Participate in public lecturers seminars

(D) Peer group interactions

18. Student bodies and groups promote friendship

19. Interpersonal relationship for personal growth
20. Interpersonal relationship for intellectual growth

665 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

21. Students helped in personal problems

22. Alcohol consumption helpful for socialization
23. Regional groups successful in socialization
23. Satisfied with socialization with peer groups

Appendix 5: Similar Questions used in survey of study by Afzal, H.,Imran,A.,

Khan,M.A. and Kashif,H. (2010)

Closed Items and Codes

The 30 items in The University Student Motivation and Satisfaction Questionnaire
Version 2 (TUSMSQ version
2) are displayed in the table, along with the item numbers, variable codes (for use in
data analysis), the target
factor, and the response scale.

No. Factor I attend university…

1 RA because I don’t know what else to do.
2 SE to understand myself better.
3 QC to gain valuable skills for my career.
4 SO because its fun place to be.
5 SP because others expect me to get a degree.
6 AL because I genuinely want to help others.
7 RA because it’s a better alternative than working.
8 SE because I want to explore new ideas.
9 QC to enhance my job prospects.
10 SO because I enjoy the social life.
11 SP because other people have told me I should.
12 AL because I want to contribute to society.
13 RA to avoid being unemployed.
14 SE because I want to challenge myself.
15 QC in order to get the qualification.
16 SO because I enjoy the social environment.
17 SP because it would disappoint other people if I didn’t.
18 AL because I want to help solve society’s problems.

PAKISTAN BUSINESS REVIEW JANUARY 2012 666

Research Predicting Factors That Affect the Student’s Academic Performance

19 RA because it gives me something to do.

20 SE for my personal growth and development.
21 QC because it will help set up my future career.
22 SO because of the social opportunities.
23 SP it seems to be the recommended thing to do.
24 AL because I want to improve the world situation.
25 RA because I don’t have any better options.
26 SE because I love learning.
27 QC so I can get a better job.
28 SO because its a great place to develop friendships.
29 SP of social expectations from those around me.
30 AL because I want to be more useful to society.

International Journal of Business and Management Vol. 5, No. 4; April 2010

The following items contain the key to sources of motivation mentioned in the
survey:-
• Rejection of Alternative options (Extrinsic) 1, 7,13,19,25
• Self-exploration (Intrinsic) 2, 8,14,20,26
• Career and Qualifications (Extrinsic) 3, 9,15,21,27
• Social enjoyment (Extrinsic) 4, 10,16,22,28
• Social Pressure (Extrinsic) 5, 11, 17, 23, 29
• Altruism (Intrinsic) 6, 12,18,24,30

667 PAKISTAN BUSINESS REVIEW JANUARY 2012

Predicting Factors That Affect the Student’s Academic Performance Research

Table 1. Sources of Motivation

Intrinsic Student with full motivation and is actually interested in learning and
Self-exploration exploring ideas for its own sake.
• Believed to do well and show good result.
• e.g. ‘I attend university because I have a genuine interest in the
subject
I am studying’.
Altruism Wants to learn for own satisfaction, becoming useful to the society,
helping others and solving their problems.
• Believe to be genuinely motivated and show better results.
• e.g. ‘I attend university because I want to be more useful to
society.’
Extrinsic Concerned with their careers and more inclined towards completing
Rejection of the degree only to get good jobs in future.
Alternative • Continuing studies just to avoid working or because does not know
Options what else to do.
• Believed to be de-motivated and can not keep a consistent
performance
academically.
• e.g. ‘I attend university because I don’t know what else I would
do’.
Career and Concerned with their careers and inclined towards completing the
Qualifications degree only to get good jobs in future.
• Concerned with getting the degree but not for learning sake.
• Believe to show good performance but not for long-term or where
the
reward does not exist anymore.
• e.g. ‘I attend university to enhance my job prospects
Social View University as a socializing place where they can have fun and
Enjoyment make friends.
• De-motivated students who are thought to show very little
academic
performance.
• e.g. ‘I attend university because I enjoy the social life’.
Social Pressure Surrounded by social pressures, peers, parents, etc.
• Try to live up to other’s expectations.
• Do not have genuine interest in studies with no consistency.
• e.g. ‘I attend university because others expect me to get a degree

PAKISTAN BUSINESS REVIEW JANUARY 2012 668