Professional Documents
Culture Documents
01 Ramzan Development 2021 2
01 Ramzan Development 2021 2
Yuncheol Kang2*
Development of a Competency Assessment
Muhammad Waqas Iqbal3** Model for Measurement of the Human
Inspection Skill
DOI: 10.5604/01.3001.0014.6074
1 National Textile University, Abstract
Department of Garment Manufacturing, Extensive research has already been done on visual inspection as well as on the effect of
Pakistan different factors on human inspection performance. However, a method should be developed
2 Ewha to measure their inspection skill based on influencing factors. This study contributes to the
Womans University,
School of Business, literature by proposing a competency assessment model based on the influencing factors
Shinsegae Bldg. #419
that can classify human labour into its respective skill levels. From the literature review, the
52, Ewhayeodae-gil, Seodaemun-gu, influencing factors of visual inspection are listed and divided into five observed variables.
Seoul 03760, South Korea A team of experts selected the significant factors with respect to the textile and clothing
* e-mail: yckang@ewha.ac.kr, industry. The analytical hierarchy process is used to measure their respective weights so
phone: +82-2-3277-4203 as to calculate the inspection performance in terms of a competency score. A numerical
GENERAL PROBLEMS OF THE FIBRE AND TEXTILE INDUSTRIES
10 Ramzan MB, Kang Y, Iqbal MW. Development of a Competency Assessment Model for Measurement of the Human Inspection Skill.
FIBRES & TEXTILES in Eastern Europe 2021; 29, 2(146): 10-19. DOI: 10.5604/01.3001.0014.6074
formance of individual inspectors as haviour, which vairies systematically for In order to reduce task complexity, fac-
well as the overall inspection station. Pi- people or groups of people of different tors such as the defect distribution, de-
oneering work was done by Harris [13] ages, genders, levels of education, and/or fect probability, defect complexity, and
on the nature of industrial inspection. cultural background [19]. Researchers number of defect types, were studied to
He presented a framework to understand considered the effects of gender and age evaluate the performance of visual search
and improve industrial inspection perfor- on visual inspection to determine the dif- and decision-making. Results showed
mance. Subsequently, much work was ference in inspection performance. How- a negative influence of defect complexity
done on visual inspection based on the ever, their studies did not find significant and a positive influence of defect prob-
results of Harris [13]. A number of factors differences in accuracy for either gender ability on the response factors [26, 27].
have been considered, and their effect on or age [19-21]. The experience of a qual- Tetteh, Jiang [28] investigated the effect
visual inspection performance has been ity inspector is also an important individ- of search strategy, task complexity, and
evaluated. The objectives were achieved ual factor that contributes positively to pacing on inspection performance. A sys-
by focusing on skills such as visual improving inspection performance. Chan tematic search strategy results in superior
search, decision making ability, and in- and Chiu [22] worked on experienced performance and decreases the inspec-
spection strategy through online and and inexperienced inspectors to inves- tion time. Moreover, task complexity
offline training [14]. In a visual search, tigate visual lobe shape characteristics was also observed as a significant factor,
inspectors carefully search for flaws, and investigate their effect on inspec- because the easier the task, the faster and
while decision-making helps to decide on tion performance. Visual lobe roundness more accurate were the inspectors. Simi-
the rejection of the item selected. On the was evident in those inspectors who had lar results were obtained by Watanapa &
other hand, inspection performance is as- long experience as compared to inexpe- Kaewkuekool [29], who worked on the
sessed on the basis of two measurements, rienced students. Visual performance de- effect of defect complexity on inspec-
inspection speed and inspection accuracy pends very much on visual capabilities, tion performance. They suggested that
[15]. Accuracy is measured in terms of hence it is always considered a suitable inspectors must be trained based upon
the hit rate, percentage of correct detec- parameter for the selection of labour various product complexities to increase
tion, and false-alarm rate, while speed can for an inspection process. Some stud- performance and save training costs.
be measured as the search time, stopping ies have been conducted to evaluate the
time, and inspection time [2]. effect of visual strength, the visual lobe When improvement in the quality of hu-
shape, and fatigue. Visual fatigue and man inspection is required, training is
Visual search is very much affected by inspection accuracy were studied to im- considered to be the primary intervention
the speed and rigidity of pacing. In terms prove inspection performance using two strategy [30, 31]. For the first time, Czaja
of accuracy, the effects of per-lot and types of wafer coatings (Nano and gold) and Drury [21] highlighted that training
per-item pacing were evaluated based and two monitor sizes (14 and 19 inch- is a neglected area when discussing im-
on inspection performance. Pacing speed es). A reduction in visual fatigue and provement in inspection performance.
proved to be a significant factor for the improvement in accuracy was observed Their results were based on detailed
accuracy of both per-item and per-lot; with a 19 inch monitor size and gold experimental reports rather than only
however, per-item is considered more coating conditions [22, 23]. general training principles. The task per-
favorable to industry [16]. The accuracy formance of three different age groups
of visual inspection is particularly impor- Other than the aforementioned factors, was observed, and it was concluded that
tant for the inspection of sensitive prod- the nature of the job and the complexity inspection errors were reduced due to
ucts, such as nuclear weapons. Recently, of a task also affect the performance of active training, while decrements in per-
see [17], visual inspection reliability was a human inspector. A pioneering study of formance due to age were also observed,
measured for precision manufactured task complexity in visual inspection was albeit smaller in magnitude. After that,
parts of nuclear weapons. Multiple in- done by Gallwey and Drury [24]. Three different types of training methods were
spections, the inspector confidence rat- types of inspection complexity were test- formulated for visual inspection, and
ing, workload, and the stress of visual in- ed based on different fault types. It was their effect on inspection performance
spection were considered to measure the concluded that inspection performance was measured [15, 28, 32]. They con-
reliability in terms of accuracy and time. is reduced due to the complexity, which cluded that a proper training program
It was also concluded that inspection is significantly affects search error, mis- based on sound principles of training
a workload intensive task dominated by judgment of fault size, and decision er- design and a well-defined methodology
mental demand and effort [17]. ror. Multitasking is another scenario that can bring significant improvements in
increases task complexity and affects inspection performance. Various training
The performance of human inspectors is inspection performance. A hybrid system methods are used for industrial inspec-
also influenced by organisational, physi- was evaluated with inspectors perform- tion, including instructional training,
cal, and individual factors [14, 18]. Or- ing a single task, three multiple tasks, online training, computer-based training
ganisational factors include the training and five multiple tasks. It was conclud- with feedback training, and feedforward
conducted, work methods, work proce- ed that multiple defect types along with training [15, 27, 28, 32-34]. Compared to
dures, policies, and social aspects. Physi- multitasking had a negative effect on offline training, online training was more
cal factors are the tools, aids, equipment, performance [25]. Similar results were valuable because it considered real world
and layout of a workplace that support also achieved by Master, Jiang [26], who situations [34]. Similarly, feedforward
the process of inspection. The individ- worked on human trust over time in hy- provides prior information regarding
ual factors are the interest, attitudes, brid systems. Their results showed that concepts, goals, and rules to inspectors
knowledge, and skills. Improvement in human trust is sensitive to the type of er- in the form of physical/verbal guidance
performance depends on the learning be- ror made by a system. and demonstration before the inspection
be used for competency assessment in terms of a numerical value. This value will help to classify
quality inspectors into their skill levels according to their inspection performance.
tors into their skill levels. This division In order to find out the significanant indi- petency assessment model as described
is based on influencing factors that cause cators of each observed variable, a struc- below.
performance variation among inspectors. tureal survey was conducted in the textile
For this purpose, we need to select the and clothing industry. 130 respondents
Application of analytical
significant indicators from Table 2 and (with a response rate of 52.0%) took part
hierarchy process
then measure the weight of all observed in the study. After anlysis, three signifi-
variables and their indicators using the cant factors from each observed variable The objective of this study is to model
multi-criteria decision making method – were selected [41]. These indicators will different influencing factors of inspec-
AHP. be used in the development of this com- tion performance into one performance
Objectiveof
Objective ofapplying
applying AHP
AHPmethodology
methodology
indicator. For this purpose, a mathemat-
Objective of applying AHP methodology ical model, the Analytical Hierarchy Pro-
cess (AHP) developed by Saaty [42], was
Selection
Selection of of expertsfrom
experts fromrelevant
relevant industries
industriesforfor
pairwise comparison
pairwise comparison used. AHP is a simple decision making
Selection of experts from relevant industries for pairwise comparison
tool that consists of decomposing com-
Selection of significant
Selection indicatorsindicators
of significant for each observed
for eachvariable by experts
observed according
variable to theaccording
by experts selected industry plex problems into components that are
Selection of significant indicators for each observed
to the variable
selected by experts according to the selected industry
industry organised into sets and then finally the
sets into levels to generate a hierarchal
Development of a hierarchy framework according to a defined objective
Development
Development of of a hierarchyframework
a hierarchy framework according
accordingtotoa defined objective
a defined objective structure [43]. The method is based on
a theory of measurement through pair-
Distribution of an assessment form for constructing judgment matrices wise comparisons relying on the judge-
Distribution
Distribution of assessment
of an an assessment form for
for constructing
constructing judgment matrices
judgment matrices
ment of experts to derive priority scales.
No These scales measure the intangibles and
No
No
Relevant literature of AHP and online training
Are experts fully aware of AHP methodology? Relevantliterature
literature ofprovided
AHP and
andonline
onlinetraining
Are experts fully aware of AHP
Are experts fully aware of AHP methodology?
Relevant wasof AHP training tangibles in relative terms using a scale
methodology? was provided
was provided
Yes of absolute judgement. It represents how
Yes
Yes
Decision process using a fundamental preference scale of absolute numbers much one element dominates another
Decision process
Decision using
process a afundamental
using preference
fundamental preference scale
scale of absolute
of absolute numbers
numbers with respect to a given attribute [44].
A complete flow chart of the AHP pro-
Pairwise comparison of criteria and calculation of their relative weights
Pairwise
Pairwise comparisonofofcriteria
comparison criteria and
andcalculation
calculationofof
their relative
their weights
relative weights cess is shown in Figure 2, and an exam-
Calculate the eigenvalue and eigenvector for all Conduct a consistency test to validate the
ple of each step as applied to this paper is
Calculate
Calculatethethe
criteria eigenvalue
eigenvalueand and
eigenvector for all
eigenvector Conduct a consistency
assessment
Conductof a test group
theconsistency
decision to validate
test to the
validate detailed below.
criteria for all criteria assessment of the decision
the assessment group
of the decision group
Definition of the objective
Pairwise comparison of the sub-criteria and calculate their relative weights The objective of this study is to develop
Pairwise
Pairwise comparisonofofthe
comparison thesub-criteria
sub-criteria and
andcalculate their
calculate relative
their weights
relative weights
Calculate the eigenvalue and eigenvector of the Conduct a consistency test to validate the
a competency assessment model based
Calculate
Conduct theaeigenvalue
sub-criteria consistencyand test
eigenvector of the
to validate Conduct a consistency
Conduct
assessment test group
a consistency
of the decision to validate
test to the
validate on the factors of visual inspection so that
sub-criteria
the assessment of the decision group assessment of the decision
the assessment group
of the decision group inspectors can be classified into their re-
spective skill levels.
Pairwise comparison for the options of the sub-criteria and their relative weights
Pairwise
Pairwise comparison
comparison forforthe
theoptions
options of the
the sub-criteria
sub-criteriaandandtheir relative
their weights
relative weights
Selection of experts for assessment
Calculate the eigenvalue & eigenvector of the Conduct a consistency test to validate the
Calculate thethe
sub-criteria’s
Calculate eigenvalue
options & eigenvector
eigenvalue of the
& eigenvector Conduct
assessmenta consistency
Conduct test group
of atheconsistency
decision to validate
test to the
validate A manufacturing sector is nominated for
sub-criteria’s
of theoptions
sub-criteria’s options assessment of the decision
the assessment group
of the decision group the selection of experts where the process
of inspection is performed by human la-
Present a final summary with the local values of the criteria, sub-criteria, and options of the sub-criteria bour. For this purpose, the value added
Present a finala summary
Present with thewith
final summary localthe
values
along ofvalues
local
with the
the criteria,
of the
global sub-criteria, and options ofand
criteria, sub-criteria,
values the options
sub-criteria
of the sub-criteria along
along with with the
the global global values
values sector of the textile industry was select-
ed, which includes garment manufactur-
The global values are used to determine the competency score of the quality inspectors (alternatives) and ing (knitwear, denim, woven, etc.) and
The global
The values
global are used
values are to determine
used thethem
toclassify competency
determine into score
their skilloflevels
the competency thescore
qualityofinspectors (alternatives)
the quality inspectorsand
classify
(alternatives) and them intothem
classify their skill
into levels
their skill levels
home textiles. A decision group of eight
Figure 2 Complete flow chart indicating the application of the AHP process people was selected with the minimum
Figure 2. Complete Figure
flow2chart
Complete flow chartthe
indicating indicating the application
application of the AHPof the process.
AHP process designation of a manager who direct-
ly looks after the process of inspection.
11 An academic researcher and an expert
Table 3. Selected indicators for each observed variable.
11
trainer of human labour from a service
Observed variables Indicators organisation accompanied this decision
School/higher school education
group. Thus, a team of experts was final-
ised that conducted the complete process
Personal factors (PF) Length of relevant experience
of AHP as described in Figure 2.
Relevant knowledge of the inspection process
Increase in the number of items coming from the manufacturing line Selection of the significant indicators
System factors (SF) Increase in the fault percentage coming from the sewing line The decision group selected three signif-
Increase in the number of defect types icant factors for each observed variable
Inspection quantity from Table 2 based on their industrial ex-
perience and the impact of the indictors
Physical /mental factors (PMF) Inspection time per item
on the human inspection skill. The re-
Inspection error
sults are summarised in Table 3.
Number of inspection tasks to be performed
Inspection task factors (ITF) More complex items to be inspected Development of the hierarchy
Inspection procedure (random or systematic) framework
Incentive system and benefits A hierarchical framework was developed,
Organizational factors (OF) Special training programs for inspectors
as shown in Figure 3. It includes the ob-
served variables, their selected indicators,
Monitoring the performance of quality inspectors
and the levels of each indicator. It is ob-
Figure 3. Hierarchy of the competency assessment model for human inspection skill.
vious from the hierarchy framework that the other sub-criteria have three options numbers (Table 4). The final membership
the model consists of five hierarchical each, as shown in Figure 3. However, the form of each comparison is then convert-
levels. The first level is the objective that last level of AHP is normally called the ed to a numerical value according to this
is referred to as the competency assess- alternatives, and in this study each quality defined scale.
ment model for the human inspection inspector is considered as an alternative.
skill. At the second level, the goal is di- The final weight of each sub-criterion’s Pairwise comparison of the criteria
vided into five main criteria or observed option will be used to assess the skill level Table 5 shows the membership functions
variables: personal factors, system fac- of the quality inspectors, i.e., the alterna- of all the elements of the criteria that
tors, physical/mental factors, inspection tive, in terms of the competency score, as include the observed variables: PF, SF,
task factors, and organisational factors. shown in Figure 3. PMF, ITF, and OF. Based on this infor-
At the third level, each criterion is then mation, a judgment matrix was devel-
divided into three sub-criteria, i.e., the se- Decision process oped that indicates the interrelationship
lected indicators by the decision group In AHP, the weights are calculated by of each observed variable taking into ac-
(Table 3). The fourth level consists of comparing each pair of criteria based on count the human inspection skill. For ex-
three options of each sub-criteria that may the assessment results that are finalised ample, experts considered PF to be three
vary with respect to each quality inspec- by the decision group. Experts are asked times more important for the human in-
tor. These options are different for each to make a pairwise comparison of all the spection skill compared to SF, and two
sub-criterion; for example, the education elements of the criteria, sub-criteria, and times more important compared to ITF.
of quality inspectors may be elementa- options of the sub-criteria using the fun- Similarly, PF is two times less important
ry, middle or high school. Similarly, all damental preference scale of the absolute compared to PMF and OF.
Table 6. Random consistency index. Source: [45]. the global values (GV) forThe each option of
resultant the sub-criteria
global values are also usingsummarised
the following equation:
(1) in Table 7 a
N 1 2 3 The 4 resultant
5 6global7 values 8 are9also 10 summarised in Table 7global
The resultant and will valuesbe used to assess
are also sum-or the skill leve
the quality inspectors working in organisations at inspectio
Random consistency index (RI) 0 0 0.52 0.89 1.11 1.25 1.35 1.40 1.45 1.49 marised in Table 7 and will be used to
the quality inspectors working in organisations or at inspection stations. In AHP, the
in- level
last cons
of an assess
alternative,the skill
which level is ofthethe quality
quality inspectors for this
The resultant global values are also spectorssummarised
working in Table 7 and will be or
in organisations used to assess t
Thus, the final judgement matrix based Then of aneigenvector
alternative,X which is is the quality
determined by inspectors
at inspectionfor this study. InThe
stations. competency
AHP, the last score for e
Then eigenvector XXHIS alternative
isisdetermined by (CSAlt)the
taking is determined
average of using
all the the following
values ininone equation
onerow of
HIS
on the human inspection skill is as fol- taking the average Then eigenvector
the quality
of all the inspectors
values in
HIS determined
working
one inlevel by taking
organisations the
consists of average
or ataninspectionof all
alternative,the values
stations.
whichIn AHP, row of
the la
alternative (CS ) is determined using the following equation:
row of the normalised . ThisNindicates
matrix HIS. This theisrelative
Alt
lows: normalised matrix NNHIS the quality weight inspectors for this present
study. in matrix A
normalised
indicates the of matrix HIS. This which
an alternative,
relative weight of indicates
each el- the relative
is the weight of
quality inspectors of eacheachthis
for
element
element
study. present in matrix A
The competency
The competency score for each alterna-
ement
Similarly, matrix Aeigenvalue
present inprinciple HIS. Similarly, λλmax, HIStive (CSAlt) is determined using the fol-
Similarly, the the principle(CS
alternative eigenvalue
) isisdetermined
isis determined
max, HIS using determined by
by the the summation
summation of of the
the products
products of of
the principle eigenvalue λmax,AltHIS deter- lowingthe following
Equation (2):equation:
mined by the summation and the of theofproducts The ultimate objective of this study is to divide the
eigenvectorXXHIS
eigenvector sum the columns of the reciprocal matrix AHIS. The following results w
HIS and the sum of the columns of the reciprocal matrix AHIS. The following results w
of the eigenvector
The ultimate XHIS and the sum
objective of ofthisthestudy is to divide the quality inspectors into their skill lev
columns of the reciprocal matrix A based
. on their performance i.e., their (2)
competency score. Be
obtained.
obtained. HIS
4.6.1. Compute the Eigenvalue and Eigenvector The following
based on theirresults were obtained:
performance i.e., their competency score. Before this, we need to define the cut-
valuesThe for the different
ultimate skill levels
objective of thisthat mustisbe
study to defined. For t
Compute the eigenvalue and eigenvector The ultimate objective divide of this the study is to divide the quality inspectors into t
quality
values for the different skill levels that must be defined. For this purpose, a normalisation process w inspectors into their
From
Fromthe
4.6.1. the matrix
Computematrix above,
theabove,
EigenvalueAAHIS
HIS , the
,and
the nor-
Eigenvector
normalised principle eigenvector, and eigenvalue are conductedskill to
computed. define
First,
levels basedthe range
on their of performance
the competency scores. F
malised principle eigenvector, and ei- based on their performance i.e.,i.e., their
their competency
competency score.
score. Before
Before this,
this,we need to de
4.6.1. Compute the Eigenvalue andwe Eigenvector conducted to define the range of the competency scores. First, the maximum and minimum val
genvalue
we need are computed.
to normalise theFirst,
matrix, forneedthe purpose of which each entry in the columnwere of the determined
matrix A using the global
we need toHISdefine the cut-off values for values of the sub-crite
to
Fromnormalise
the matrixthe matrix,
above, fornormalised
AHIS, the the purpose principle eigenvector, and values for the
eigenvalue are different
computed. skill
First, levels that must be defined. For this purpose, a normalisat
were determined using the global values the of the different skill levels
sub-criteria optionsthat must bein Table 7. T
mentioned
ofiswhich
divided each
by entry
the sum in of
theitscolumn
respective of the column.4.6.2. Consistency
The normalised
4.6.2. Consistency matrix Test
TestNHIS was developed maximum valueForwas
as follows:
defined. this0.60 and thea normalisa-
purpose, minimum value 0.10.
Consistency and test
conducted aretocomputed.
From
we need
matrix theA
tomatrix
HIS isabove,
normalise the
dividedAHIS,by
the the
matrix, normalised
for sum ofprinciple
the purpose eigenvector,
its of which each entry
maximum
eigenvalue
in thewas
value column 0.60ofdefine
the
andmatrix
the
First,range
the Aminimum
HIS
of the competency scores. First, the maximum and m
tionvalue
process wasThen,
0.10. conducted to define the
the maximum values (0.60) w
respective column. The normalised ma- After calculating the eigenvalue λ max, HIS,considered asofthe normalisation constant, and both the maxi
were determined using the range
global values the of competency
the sub-criteria scores. First,
options mentioned in
we
trixneed
NHIStoby
is divided normalise
was sumthe
thedeveloped matrix,
of its as for
respective thecolumn.
follows: purposeThe of which
and each
After
normalised
After
considered
entry
eigenvector
matrixin Nthe
calculating
calculating Xcolumn
thewas
as HIS the
the
HIS of the matrix
, developed
a consistency
eigenvalue
eigenvalue
normalisation λAλmax,
as follows:test
HIS
constant, HIS, and
and
max, HIS, and theboth
eigenvector
eigenvector
maximum the maximumXXHIS
and
, , aaand
minimum
HIS
consistency
consistency
minimum test
test
values were
was
waswere
values conducted
conducte
divid
was conducted to verify the assessment by this normalisation constant to define the range of the com
is divided by the sum of its respective column. The normalised matrix Nmaximumwas developed value was
asdecision
follows:0.60group.anddetermined
the minimum usingvalue the 0.10.globalThen, values
CI the of maximum valu
verify
ofby the the
decisionassessment 0.17CR
HIS group. of
For thethis purpose, For this purpose, the
verify
this normalisation constant to definegroup.
the assessment of the decision the the Forsub-criteria
range this
of thepurpose,
competency thevalue
options value
scoreof CIand
ofi.e.,
mentioned and CR
in to was ofcalculat
1was calcula
the mo
the value of CI and CR was calculated, 16
considered as the normalisation Table
constant, 7. The maximum
and both the value wasand
maximum 0.60minimum value
the
the outcomes
outcomesofof
outcomes
the which
ofwhich
whichare are presented
arepresented
presentedbe- below.
below.and the minimum 16 value 0.10. Then, the
low:
by this normalisation constant tomaximum define thevalues range (0.60)
of the were considered
competency score i.e., 0.17 to
as the normalisation constant, and both
the maximum and minimum 16 values were
divided by this normalisation constant to
define the range of the competency score
Then eigenvector XHIS is determined by taking the average of all the values in one i.e.,row 0.17of the
to 1 of the model proposed.
Where the value of RI for the 5×5 matrix Based on 15 this
15 range, the decision group
normalised matrix NHIS. This indicates theisrelative 1.12, asweight
mentioned of each in element
Table 6, presentand n indecided matrix Athe HIS. cut-off values of the three
Then eigenvector XHIS is determined by takingindicates the averagetheofnumber
all the values in one row
of elements of the
present different skill levels:
Similarly, the principle eigenvalue λmax, HIS is determined by the summation of the products of the
Thenmatrix
normalised eigenvector XHIS isindicates
NHIS. This determined the by takingweight
relative the average of allelement
of each the values in onein row
present of the
matrix AHIS.
16 FIBRES & TEXTILES in Eastern Europe 2021, Vol. 29, 2(146)
eigenvector XHIS and the sum of the columns of the reciprocal matrix AHIS. The following results were
normalised
Similarly, thematrix NHIS. This
principle indicates
eigenvalue λmax,theHISrelative weight of
is determined by each element present
the summation matrix Aof
of theinproducts HIS.the
obtained.
Similarly, the principle eigenvalue λ is determined by the summation of the products of the
remove any ambiguity. Finally, the data generated are used to calculate the competency score of all
the quality inspectors using the global values. Figure 4 shows the competency score of all quality
Competency score
Competency score
score was greater than 0.50 and less 0.60
0.60
than or equal to 0.75
0.50
0.50
3. For high skill, the competency score
was greater than 0.75 0.40
0.40
0.30
0.30
Then, the global values were used to de- Education 0.16 Basic 0.32 0.010
Above tobasic 0.56 results
0.018
termine the competency score for each However, it was observed that the cut-off values needed be redefined, as the of the
Personal factors
respective options. After that, the refine- One type of defect 0.06 0.004
ment process of random data is done by
Competency assessment model (CAM) for human inspection skill (HIS)
quantity
from 0.27 to 0.88. Above 80% of Target 0.59 0.033
Above 60% more than ST 0.08 0.010
The next step is to divide the quality in- Inspection
0.34 0.39 31 to 60% more than ST 0.26 0.034
time
spectors into their skill levels based on 1 to 30% more than ST 0.66 0.085
their competency score. Accordingly, Less than 5% 0.67 0.099
Inspection
the aforementioned cut-off values of the error
0.44 6 % to 10% 0.24 0.036
three different skill levels in the previous Above 10% 0.09 0.013
section for low, medium and high skill Scanning 0.12 0.002
Search
quality inspectors were 318, 646, and 36, 0.20 Measuring 0.32 0.006
Inspection task factors
strategy
respectively. However, it was observed Both 0.56 0.011
that the cut-off values needed to be re- Basic product 0.17 0.008
Inspection
defined, as the results of the numerical 0.10
tasks
0.49 Complex product 0.39 0.019
example were not realistic as far as the Highly complex product 0.44 0.021
number of high skill inspectors were con- Random 0.15 0.005
Product
cerned. For this purpose, first a normality complexity
0.31 Symmetric 0.38 0.012
test was conducted using IBP-SPSS22. Both 0.47 0.014
The test results (Table 8) were signifi- No incentive 0.09 0.010
Incentive
cant, and the competency score for all 0.42 Reward only 0.29 0.031
systems
Organisational factors
the quality inspectors based on randomly Reward and punishment 0.62 0.065
2) For medium skill, the competency score was greater than 0.45 and less than or equal to 0.65
3) For high skill, the competency score was greater than 0.65
system with more focus on measuring
33 Medium
Medium skill
skill the skills of labour and classify them
into their respective levels. Previous
2.5
2.5 studies have evaluated the effect of dif-
ferent factors on inspection performance.
22 However, significant variables were not
modelled to fully measure the inspection
1.5
1.5
performance in terms of skill levels. This
study identified the influencing factors
and utilised them to design a scale. In this
11
regard, a competency assessment model
was proposed to determine the score of
0.5
0.5
Low skill
skill High skill the inspectors based on their performance
using AHP. The model proposed was ap-
00
plied to randomly generated data of the
0.00
0.00 0.20
0.20 0.40 0.60
0.60 0.80
0.80 1.00
1.00
inspectors and their competency score
was measured successfully. The results
Figure 5. Normally distributed data of the competency score. helped to define the cut off values for the
Figure 5 Normally distributed data of the competency score three skill levels of inspectors, i.e., low,
According
Table toofthe
8. Results therevised
normalitycut-off values,
test. Note: a the quality
Lilliefors inspectors
significance mediuminto
were again classified
correction. and high
theirskill and all the inspec-
tors were classified into their skill levels
Kolmogorov-Smirnova Shapiro-Wilk based on their competency score.
respective skill levels. Then, the comparison was summarised for the different cut-off values, shown
Statistic df Sig. Statistic df Sig.
Competency score 0.036 1000 0.004 0.997 1000 0.044 Firstly, the model proposed provided the
in Table 9. This classification seems to be realistic and therefore the revised cut-off values
most canfactors
effective be that should be mon-
itored to get the maximum output from
used forTable 9. Number of quality
the classification inspectors
of quality based on or
inspectors theany
initial and revised cut-off
organisation values.station.
or offline the inspectors. Secondly, the competency
According to the initial cut-off values According to the revised cut-off values assessment enables the organisation to
Low skill Medium skill High skill Low skill Medium skill High skill measure the inspection skill in the form
Cut-off values 0.17-0.50 0.50-0.75 0.75 0.20-0.45 0.45-0.65 0.62 of a competency score, providing a basis
No. of inspectors 318 646 36 165 663 172 to rank the available inspectors according
19 to their respective skill levels based on
objective data. In this way, managers can
ues of the competency score,whic ranged the model proposed has successfully efficiently utilise their manpower accord-
from 0.27 to 0.88, the experts assumed measured the human inspection skill by ing to its working capacity. It also devel-
a range from 0.20 to 0.90 because it determining the competency score using ops an atmosphere of competition among
might be very rare that the final compe- randomly generated data. The resultant the labour in which every quality inspec-
tency score would be lower than 0.20 or skill level is based on the influencing tor will be motivated to improve his or
higher than 0.90. Thus, the revised cut- factors of the visual inspection that can her competency score by improving their
off values are described as follows: affect the performance of human labour. inspection performance. Since the model
1. for low skill, the competency score The numerical example also helped to proposed is based on a comprehensive
was greater than 0.20 and less than or redefine the cut-off values of the three framework of influencing factors, it en-
equal to 0.45, different skill levels based on the results. ables mangers to focus on deficient areas
Thus, the model proposed is capable of that cause the low performance of an in-
2. for medium skill, the competency
determining the competency score and dividual quality inspector. In conclusion,
score was greater than 0.45 and less
can be used to classify the human labour this research supports the idea that clas-
than or equal to 0.65,
of an inspection station into its respective sifying human labour into its skill levels
3. for high skill, the competency score
skill levels. In this way, an organisation is more important for organisations to
was greater than 0.65.
will be able to utilise its manpower ac- improve and achieve overall efficiency,
cording to its performance capacity. It because the workforce will be utilised
According to the revised cut-off values,
will also help to develop a pay scale and rewarded according to its abilities
the quality inspectors were again clas-
for human labour based on its compe- and skills. However, this aspect needs to
sified into their respective skill levels.
tency score. In conclusion, this method be focused on further and practitioners
Then, the comparison was summarised
proposed will create an atmosphere of should work to improve the skill levels
for the different cut-off values, shown in
competition among human labour that of inspectors so that optimal results can
Table 9. This classification seems to be
will improve the individual and overall be obtained from available manpower.
realistic and therefore the revised cut-off
inspection performance. It is also recommended that micro level
values can be used for the classification
studies be conducted in which each ob-
of quality inspectors or any organisation
or offline station. served variable and its respective indica-
Conclusions tors must be investigated to evaluate their
It can be concluded from the results of The primary objective of this research effect on inspection performance.
the numerical examples presented that was to study a human based inspection