You are on page 1of 6

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/267932780

Students" Performance Evaluation Using Statistical Quality Control

Article · January 2012

CITATIONS READS

5 2,461

1 author:

Birhanu Beshah
Addis Ababa University
48 PUBLICATIONS   325 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Technological Capability Development View project

Layout Design View project

All content following this page was uploaded by Birhanu Beshah on 08 November 2014.

The user has requested enhancement of the downloaded file.


International Journal of Science and Advanced Technology (ISSN 2221-8386) Volume 2 No 12 December 2012
http://www.ijsat.com

Students‟ Performance Evaluation Using


Statistical Quality Control
Birhanu Beshah
School of Mechanical and Industrial Engineering
AAiT, AAU
Addis Ababa, Ethiopia
bireabesha@yahoo.com

Abstract—One output of a well-functioning educational II. BRIEF DESCRIPTION ABOUT THE PROBLEM
system is knowledgeable students. In higher education,
evaluating and improving student performance can be The number of universities in Ethiopia and their intake
challenging. This study applies Statistical Quality Control capacity has increased dramatically in recent years
(SQC) to help address this challenge. Using randomly selected
(Government of Ethiopia, 2008). With the vision of
student scores from 20 sections of Addis Ababa Institute of
Technology’s Engineering Mechanics II course, the paper transforming agriculture-based economy into industrial-
uses statistical quality control to develop a new grading based economy, the system of education now focuses in a
method. The new approach compares favourably with the proportion of 70% with science and technology education
prior method of grade determination. In addition, the SQC- and 30% arts and social sciences. Technology and
based approach may allow faculty to give grades more engineering constitute 40% of the total enrolment in higher
objectively than before and students likely will perceive the education. In the last four years, the number of engineering
new approach as more fair and, therefore, more motivating. students in Addis Ababa Institute of Technology (AAiT),
first and the largest engineering and technology institute in
Keywords-Grading system; student performance
the country has increased more than tenfold [10]. In
evaluation
parallel, however, the attrition increased at a very high rate.
This becomes major concern for the Ethiopian Ministry of
I. INTRODUCTION Education and Addis Ababa University.
After the development of statistical Quality Control (SQC)
by Shewhart [16] quality specialists such as, Deming, After interviewing AAiT‟s students and instructors, higher
Crosby, Juran, Feigenbaum, Garvin, Ishikawa, Taguchi and officials from the Ministry of Education and Addis Ababa
Oakland played a very important role in the application and University held a meeting on May 26, 2012 to discuss the
dissemination of SQC concept across all industries problem with AAiT‟s top and middle management. The
categories. The uses of SQC in education has been clearly Scientific Director, department heads and chairs and the
seen in the area of students evaluations of teaching [9], writer of this article were present. Background of the
student grade monitoring [3], grading process in the higher students, instruction methods, facilities and laboratories,
education [14], large-scale assessment programs [15] and class-size, evaluation system, teaching materials, etc. were
assessing learning outcomes [2]. These research activities cited as the reasons for the high level attrition rate in AAiT.
were focused on the classical application of SQC: On December 18, 2012, the AAU president and academic
differentiating and identifying the assignable causes from vice president held a meeting with the AAiT academic staff
that of the random causes to continuously improve a members on the implementation of modular course
system. delivery, student evaluation and grading mechanism that the
university consider as a strategy to improve quality of
However, up to the knowledge of the researcher, this education. The new guideline issued on December 3, 2012
application of SQC to evaluate performance of students in put the approach to convert raw mark into letter grade
higher education has not been established up to now. In which bases criteria reference (fixed grading scale) (See
particular, this study, deals with the means and methods of Table1). The criteria set in the new method is an arbitrary
changing students‟ Quantitative (numeric) Results (QR) cut-off points and it will enforce instructors to give different
into Qualitative (letter) Grades (QG). It should be noted that grades for those who do not have difference thereby the
improving teaching quality in general is beyond the scope current state demotivate students. In this paper students‟
of this research. performance evaluation and means of grading students is
dealt with as follows.
Problems associated with students‟ performance evaluation
in the Addis Ababa Institute of Technology are presented in
brief. Actual data are collected and analysed to understand
the current phenomena and ultimately to propose alternative
way of grading students.

75
International Journal of Science and Advanced Technology (ISSN 2221-8386) Volume 2 No 12 December 2012
http://www.ijsat.com

TABLE I. FIXED GRADING SCALE OF AAU students‟ size was small, individual measurements were
Raw Mark Interval of Fixed Letter used to develop an X chart and the moving average chart to
Interval Number Number Grade
[100%] Grade Grades label students‟ performance by a letter grade. It would
[95, 100] 4 4.0 A+ determine students‟ performance. After, the new system
[90, 95) 3.5 ≤ X < 4.0 3.7 A was then deployed to generate QG from the students‟ QR.
[85, 90) 3.0 ≤ X < 3.5 3.3 A- Finally, those grades that means: given by the old method
[83, 85) 2.7 ≤ X < 3.0 3.0 B+
[80, 83) 2.5 ≤ X < 2.7 2.7 B and generated by the control charts were compared in
[78, 80) 2.3 ≤ X < 2.5 2.5 B- randomly selected group to justify the difference.
[70, 78) 2.0 ≤ X < 2.3 2.3 C+
[60, 70) 1.7 ≤ X < 2.0 2.0 C
[50, 60) 1.3 ≤ X < 1.7 1.7 C-
IV. X BAR AND R CHARTS
[40, 50) 1.0 ≤ X < 1.3 1.3 D Quality characteristics can be evaluated using either an
[30, 40) = 1.0 1.0 F (take supplementary attribute or variable measures. Thus, there are two types of
exam)
[0, 30) 0 0 F (repeat the course) control charts: Control charts for variables and Control
charts for attributes [13]. Control charts for variables
measurements in a continuous scale such as length, weight,
In higher education, representing students‟ performance volume, pressure, temperature or time. Moreover, Control
qualitatively using letter grades is common in many charts for variables summarize the output of a process, or
universities of the world. However, recently different operation over time whereas attribute data have only two
approaches being deployed to change QR into QG. For values such as good/bad, conforming/non-conforming, or
instance, some universities use fixed scale such as the acceptable/not acceptable. Among these charts, X bar and R
Addis Ababa University while others may do it arbitrarily chart are the most applicable and commonly used Control
based on the instructors‟ decision. For engineering Charts for variables [7, 12]
education, where the examinations are subjective and vary
considerably between batches and instructors, using a fixed The X-bar chart receives its inputs as mean of a sample
scale is irrational and most of the time, the evaluation of taken from the process under study. Usually, the sample
students would be discouraging and demotivating case. In will contain four or five observations. X-bar chart is a
the second approach, the students‟ evaluation is entirely control chart used to monitor changes in the mean value of
dependant on the instructor‟s judgment: often skewed a process whereas R chart is a control chart that monitors
negatively. Both these methods have limitations and would changes in the dispersion or variability of a process. Both
have a tendency to retard the development of education and charts have similar construction methods and they consist
knowledge. This phenomenon necessitates finding a new of a Central Line (CL), Upper Control Limit (UCL) and
method of students‟ performance evaluation in higher Lower Control Limits (LCL). Furthermore, they may have
education. also one sigma and two sigma limits on both sides of the
CL. The formula to calculate these values are represented
The general use of statistics in determining grades seems a by equation (1); (2); (3); (4); (5) and (6).
common practice in many universities. Many instructors
determine the mean and standard deviation for individual
assignments and final grades and use these to help assign X bar chart
grades. These techniques to give grades propose the use of 
statistical process control charts to help in equating levels of CL = X (1)
student performance with letter grades.
 

III. DATA COLLECTED AND THEIR PURPOSE UCL= X  A2 R (2)


Among many courses given in AAiT, Engineering  
Mechanics II (Dynamics) students‟ performances were LCL = X  A2 R (3)
studied considering both QR and QG because the statistical
trend showed that Dynamics is one of the courses where
many students failed in the pre-engineering program. From R chart
the 2010 entry, twenty groups‟ Dynamics course result was 
collected from the instructors. Each group had a class size CL= R (4)
of about 40 students. For the purpose of statistical analysis, 
the results of five students from each class were randomly UCL = D4 R (5)
selected.

The data collected were analysed and interpreted by using LCL = D3 R (6)
X bar and R charts. These charts were chosen because the
study is being carried out to measure students‟ performance Where; = mean, A2, D3, and D4 are constants and found
of a very large size, in this case, twenty. Each section has its in table. For a sample size, n=5, such as data used in this
own instructor, and the overall teaching environment of one article, A2 = 0.577, D3 = 0 and D4 = 2.115
section is different from the other sections. Hence, sample
of students were taken to find the mean and, it is due to this
reason that X bar and R charts were used. When the
76
International Journal of Science and Advanced Technology (ISSN 2221-8386) Volume 2 No 12 December 2012
http://www.ijsat.com

V. THEORETICAL FRAMEWORK
Where the performances of students in a class follow a UCL „A‟ Region
normal distribution their results should also follow normal 3σ
distribution. In spite of the consensus on students‟ „B‟ Region

Average
performance normality, setting intervals to reach at 1σ
normally distributed grade was a problem. In order to have CL
„C‟ Region
a normal students‟ grade the different regions of the control
charts are classified into six parts and each region is ─1σ
designated to a specific grade according to the students‟ „D‟ Region
LCL
results and the overall population. ─3σ
„F‟ Region

Results which fall outside the UCL, since, they are


Subgroup
representation of excellent and outstanding performer
Figure 1. New grading system proposal
beyond the system, they deserve „A‟ grade. The system has
to study these students to learn from their achievements. VI. APPLICATION OF X BAR CHART
Those which fall below the LCL, since their performance is The marks of the group, range and average are reported in
exceptionally weak, have to be advised to repeat the course Table2. Based on the data, first, R chart is plotted to check
because they have failed having scored an „F‟ grade. These the variability of the study unit and in this case all the
group are not the system result, their result is weak, may be, points were in- control which validates the X bar chart (see
due to personal reasons. Hence, the students have to be Figure2A). Then after, the X bar chart is drawn. The
approached and studied to solve their specific personal intervals from the X bar chart are used to determine the
problems. students‟ grade (see Figure2B). According to the findings
from the chart, the grades intervals for „A‟ become above
Region between the UCL and the LCL is the result of the 68.4, „B‟ between 51 and 68.4, „C‟ between 35 and 51, „D‟
between 17.6 and 35, and „F‟ below 17.6. However, the old
educational system. Contrary to specifications in
method‟s intervals for „A‟ was above 74, „B‟ between 59
manufacturing industries where the CL is the target value, and 74, „C‟ between 39 and 59, „D‟ between 29 and 39, and
in education the target value or the desired value is always „F‟ below 29. When these two methods are compared, the
higher value. Due to this, the regions in limits are classified old method grade intervals were higher than the new one.
into three equal regions and each region is assumed to have Due to this, the students‟ grade was found to be lower and
different grades: ‘B’ between +3σ and +1σ limits, ‘ the attrition rate becomes very high. It does not intend that
C’ between +1σ and ─1σ limits and ‘D’ between ─ the new method is to increase the student marks simply by
1σ and ─3σ limits. Readers may ask as to why we give lowering the cut-off. Rather, the conclusion it is that when
different label while the results in the three sigma limits are the students‟ performance is the result of the system they
the outcome of the education system. In the manufacturing need not be penalized but a means must be found to
industry one and two sigma limits are known to measure improve the system.
TABLE II. SAMPLE DATA
variability of a system. As stated above, in education, the Sub Sample Number
target is the above control limit and hence the value of the
Groups 1 2 3 4 5 Range Average
upper control limit and the lower control limit will have
significant difference in their performance. Due to these G1 90 60 64 57 42 48 62.6
two reasons, the three sigma limits are supposed to be G2 17 24 21 50 33 33 29
categorized into three distinct fields. Administrators and G3 68 27 47 28 36 41 41.2
instructors should always put effort to reduce the variation G4 18 36 50 90 21 72 43
and move the system average up to a higher level. In effect, G5 43 78 30 21 50 57 44.4
continuous improvement initiatives should be put. Figure1 G6 33 35 37 20 59 30 36.8
shows the different regions of the control charts and their
G7 22 48 30 18 100 78 43.6
respective grades.
G8 41 35 26 22 13 28 27.4

Therefore, the hypothesis of this research is that “different G9 20 33 21 28 16 17 23.6


regions of X bar chart could specify the QGs‟ intervals in G10 32 42 49 25 63 38 42.2
higher education students‟ performance evaluation (inspired G11 51 20 51 19 44 32 37
by the work of Burney [1])”. G12 35 35 21 25 76 55 38.4
G13 85 29 64 55 55 56 57.6
G14 38 68 42 55 40 20 48.6
G15 55 50 47 33 43 22 45.6
G16 70 84 15 23 72 69 52.8
G17 47 29 65 22 48 43 42.2
G18 49 40 33 51 23 28 39.2
G19 53 32 50 34 62 30 46.2
G20 87 51 58 73 26 61 59

77
International Journal of Science and Advanced Technology (ISSN 2221-8386) Volume 2 No 12 December 2012
http://www.ijsat.com

VIII. CONCLUSION
The application of SQC as an industrial problem-solving
100. UCL 93.1 tool date back to the beginning of the 20th Century since
then, they have been used widely including the service
sectors [11]. This study focused on the application of SQC
Range

50. CL 44.1 in the education sector. In fact, according to the literature


review made on non-standard application of SQC by [8],
SQC in education is not also new, but what makes this
0. research different and perhaps the first of its kind is its use
1 2 3 4 5 6 7 8 9 1011121314151617181920 to evaluate performance of students in higher education.
Subgroup
A. R chart Higher education institutions are where many people start
there life-long career and, at the same time they are places
where many people‟s potential intelligence could be
89. wasted due to improper evaluation. Limitation in this
UCL regard is the grading scheme adopted by instructors which
69. 68.4
considerably varies and often does not favour students, as
Average

51 it ascribes low performance of the system to the student‟s


49.
ability or competence.
35
29.
LCL 17.6 In general, from the academic point of view, when the
9. students‟ performance as a whole is low, below a certain
1 2 3 4 5 6 7 8 9 1011121314151617181920 limit, it‟s common to give least grade in order to maintain
Subgroup quality of education. Although students‟ performance is the
result of the educational system to a certain degree which
B. X bar chart
should be examined and revised putting the blame entirely
on the students and penalizing them was the common
Figure 2. X bar and R charts
practice over the past years. In one way or the other
students‟ failure, explains the weakness of the educators as
VII. NEW AND OLD METHOD well.
Based on the scale defined in the above X bar chart, a new
grade is given for randomly selected group; these grades This study has attempted to device a new method of
are compared with the grades given by the institute‟s changing students‟ QR into QG. In addition, it has been
instructors i.e. the old method of grading system. For proofed that the new grading system is more reliable and
further statistical analysis, the grades A, B, C, D, and F are practical than the traditional counter part. Researchers
converted into a scale as 5, 4, 3, 2, and 1 respectively for could further test this theory in different circumstances and
both systems. Furthermore, the normal distribution of the at different levels to diversify and enhance SQC‟s
two grading system (with a scale 1-5) is plotted as shown application to students‟ performance evaluation.
below. Figure3 proofed that the traditional and the new
methods have remarkable variations. The mean of the REFERENCES
traditional method grade distribution is negatively skewed [1]. Burney, A. Farhat. and Al-Darrab, Ibrahim, “Performance
whereas that of the new method was found to be a perfect Evaluation using Statistical Quality Control Techniques”, Work
Study, Vol. 47, No. 6, pp. 204-212, 1998
normal; “C” grade is the population mean. Moreover, the
standard deviation of the new method is found to be lower [2]. Cervetti, J. Michael., Royne B. Marla. and Shaffer, J. Michael,
which guarantees the reliability and consistency of the new “The Use of Performance Control Charts in Business Schools: A
method. Tool for Assessing Learning Outcomes”, Journal of Education for
Business, 87: 247–252, 2012

0.4
[3]. Deming, Edwards, “Out of the Crisis”, MIT, 2005

[4]. Edwards, H. P., Govindaraju, K. & Lai, C. D, “A Control Chart


0.3 Procedure for Student Grade Monitoring”, Res. Lett. Inf. Math.
Variable
Traditional Method
Sci., Vol. 7, pp 117-126, 2005
New Method
Density

0.2 Mean StDev N [5]. Government of Ethiopia, “Technology and Technology Capacity
2.756 1.113 41
3.098 1.091 41
Building Development Directions”, Addis Ababa, Ethiopia, 2008

0.1 [6]. Ishikaw, Kaoru, “What is total quality control? – The Japanese
way”, Printice-Hall Inc., 1985
0.0
1 2 3 4 5 [7]. Kitaw, Daniel, “Industrial Management & Engineering Economy:
Students' Grade An Introduction to Industrial Engineering”, AAU Press, 2009

Figure 3. Normal distribution of students‟ grade given by the new and [8]. MacCarthy, B.L. and Wasusri, Thananya, “A Review of Non-
traditional methods standard Applications of Statstical Process Control (SPC) Charts”,

78
International Journal of Science and Advanced Technology (ISSN 2221-8386) Volume 2 No 12 December 2012
http://www.ijsat.com

International Journal of Quality and Reliability Management, Vol.


19, No. 3, pp. 295-320, 2002

[9]. Marks, B. Neil. and O‟Cornnell, T. Richard, “Using Statistical


Control Charts to Analyse Data from Students Evaluations of
Teaching”, Decision, Sciences Journal of Innovative Education,
Vol. 1, No. 2, pp. 259-272, 2003

[10]. Ministry of Education, “Annual Intake and Enrollment Growths


and Professional and Programm Mix of Ethiopian Public Higher
Education: Strategy and Conversion Plan”, Addis Ababa, Ethiopia,
2008

[11]. Mohanty, R.P. and Lakhe, R. R, “TQM in the Service Sector”,


Jaico Publishing House, Mumbai, 2002

[12]. Oakland, John S, “Statistical Process Control”, 6th edition,


Butterworth-Heinemann publications, 2008

[13]. Reid, R. Dan and Sanders, Nada R, “Operations Management: An


Integrated Approach”, 3rd edition, John Wiley & Sons, Inc., 2007

[14]. Savić, Mirko, “p-Charts in the Quality Control of the Grading


Process in the High Education”, Panoeconomicus, Vol. 3, pp. 335-
347, 2006

[15]. Schafer, D. William., Coverdale, J. Bradley., Luxenberg, Harlan.,


Jin, Ying, “Quality Control Charts in Large-Scale Assessment
Programs”, Practical Assessment, Research & Evaluation, Volume
16, Number 15, 2011

[16]. Shewhart, W.A, “Economic Control of Quality of Manufactured


Product”, American Society for Quality Control, 1980

79

View publication stats

You might also like