*Corresponding Author, e-mail: aal[email protected]u.sa
GU J Sci, Part A, 8(3): 318-326 (2021)
Gazi University
Journal of Science
PART A: ENGINEERING AND INNOVATION
http://dergipark.gov.tr/gujsa
Araştırma Makalesi
Research Article
Designing an Effective Questionnaire using Orthogonal Array for Improving
Inventory Accuracy at Electricity Warehouse
Abdulaziz S. ALKABAA
1*
, Osman TAYLAN
1
1
Department of Industrial Engineering, Faculty of Engineering, King Abdulaziz University, P.O. Box 80204, Jeddah 21589, Saudi Arabia
Keywords
Abstract
Orthogonal Array
Stated-Choice Survey
Discrete Choice
Experiments and
Analysis
Inventory Control
Inventory Accuracy
The process of classical data collection through a survey is time-consuming, expensive, and may not be
feasible. In practice, it is always good to make sure that the survey design works with the diversity of
factors and levels that affect the products or services. Since the data quality gathered depends solely on
the quality of the survey constructed and the written questions, this paper presents one of the most
effective tools in designing a stated-choice survey using orthogonal array design to gather the
information that improves the inventory accuracy in electricity company’s warehouses. The stated-
choice survey provides an effective way to explore survey respondents’ preferences and evaluate the
interchanges that survey respondents make in considering several product and service combinations.
This method will help inventory engineers and workers for data collection with more information and a
minimum number of questions and efforts.
Cite
Alkabaa, A. S., & Taylan, O. (2021). Designing an Effective Questionnaire using Orthogonal Array for Improving Inventory
Accuracy at Electricity Warehouse. GU J Sci, Part A, 8(3), 318-326.
Author ID (ORCID Number)
Article Process
A. S. Alkabaa, 0000-0001-9016-4241
O. Taylan, 0000-0002-5806-3237
Submission Date
Revision Date
Accepted Date
Published Date
1. INTRODUCTION
Discrete choice experiments and analysis (DCA) are widely used in marketing research, health economics, and
other areas. They have been used to establish policy or survey questions, construct consumer preferences on a
specific topic, and assess current situations. Besides, they are instrumental when exploring and understanding
people’s preferences when data are not available of alternative products and services. In choice experiments,
respondents are offered a set of different choices to be selected from a set of possible options.
In logistic, there are several criteria evaluated and considered simultaneously before purchasing a material, or
a service. Verma & Thompson (1996) stated that purchasing decision about a material or service might
consider several parameters such as service or product quality, speed of delivery, special incentives for buying,
and price. Managers want to know how consumers integrate, value and exchange different product and service
features. The DCA approach provides a robust and systematic way of recognizing the relative weights and
change in attributes caused by decision makers' preferences. DCA approach is based on well tested, and
relatively comprehensive random-utility theory (RUT) leading to a wide-range of traceable and testable models
of selection behavior. In fact, RUT delivers a theoretical link among surveys, experiments, and the other forms
of stated choices and the actions detected in real-life settings. Categorical outcomes are obtained from discrete
choices because they select one alternative from each set of alternatives. Similarly, mathematical models
demonstrating customer choice of product and services can be linked to budgets, labor scheduling, service
offerings and several other factors called operating decisions. The statistical models can be employed from
discrete-choices of buying processes and incorporated in decision-support systems (DSSs) to build a model
319
Abdulaziz S. ALKABAA, Osman TAYLAN
GU J Sci, Part A, 8(3): 318-326 (2021)
for unique approaches. Dolnicar (2013) offered certain schemes for measurement in the social sciences to draw
clear conclusions about various measurement challenges. Developing good survey questions is crucially
important and precisely depend on defining what is being measured. For most measurements, experts argue
for the requests of a good definition of the goals, however unable to provide helpful guidelines as how to build
the definition that conforms with the criteria stated. Theoretical meaning of factors is usually discussed, for
instance Bagozzi (2011) stated the ambiguity, vagueness, of factors. Dolnicar (2013) argues that good
definition of criteria had better not be vague, imprecise, ambiguous, or unclear. The criteria must be obviously
understood by the respondents who should agree on their meaning. When service marks are considered, if net
definition is not allowed, will increase the uncertainty of decision-making in providing information. This is
considered as the uncertain information, for reducing the effects of uncertain information, the intuitionistic
fuzzy sets can be presented by considering the membership degrees (Mi et al., 2019) to reduce the ambiguity
and vagueness. Yager (2017) presented the intuitionistic fuzzy sets to overcome the dilemma and develops the
degree of flexibility in representing the information preferred. Vetschera (2017) stated that in decision-making
problems determining the weights of criteria is the primary concern. In this context, analytic hierarchy process
(AHP) is an appropriate method Abdi & William (2010) and Ho & Ma (2018) for weight-determination like
fuzzy AHP, fuzzy TOPSIS and fuzzy VIKOR approaches. The economic globalization brought economic
integration of the world, the growth of services continue that becomes fast and steadily, is a global strategy of
comprehensive industries (Meimand et al., 2017). For instance, VIKOR and TOPSIS approaches are practical
methods to solve decision making problems, specifically in the condition where the assessment of options over
buying criteria are conflicting. The fuzzy extensions of TOPSIS, VIKOR and AHP play an important role in
the intuitionistic and hesitant fuzzy information, and linguistic information with hesitation (Liao et al., 2015)
for solving the decision-making problem.
On the other hand, the qualitative research approaches can be carried out for the intent of apprising the
enhancement of questionnaires for the inventory management solely. Nowadays, many surveys are conducted
online with the questionnaires time taking 15 min on average to complete it. Surveys taking longer than 15
min might need less randomness or include systematic errors related with fatigue or boredom (Hoerger, 2010;
Brosnan et al., 2018). Fan & Yan (2010) investigated short surveys and found out that they are less expensive
and have several advantages over long surveys, they may return higher completion rates. Brosnan et al. (2018)
proposed a statistical method for the evaluation and analysis of surveys, which is built on principal components
analysis (PCA) for the elimination of redundant information and items with the loss of minimum information
and optimal information gain. Abdi & Williams (2010) also studied the PCA to solve problems for determining
the principal components. Yet, traditional PCA is not able to recognize the differences of items redundant and
to provide optimum information for survey items. PCA suggests potential for reducing survey superfluous
items. Brosnan et al. (2018) stated that no item is superfluous if each major component exhibits significant
variability. Specialized sampling techniques are employed to attain balance between costs and statistical
accuracy. Kolenikov (2010) stated that data collection in large-scale statistical surveys involves data clustering,
data stratification, multiple and unequal selection possibilities, and sampling with or without displacement.
Survey-sample designers use stratification to increase efficiency, protect against poorly unbalanced samples,
to optimize the overall survey cost. Sampling weights are used for guaranteeing the unbiased estimation which
are adjusted so that the weighted totals of poststratification variables fit the population totals. Nevertheless, a
precise accounting for a complex survey design is extremely cumbersome.
The statistical method aims to decrease survey size, to solve the managerial problems of acquiring optimal
information with minimal cost involving data collection across multiple points in time to improve data quality,
both longitudinal and repeat cross-sectional. The PCA is a standard method for reducing a set of survey items
(Boyes et al., 2009). Cervellera et al. (2007) presented multivariate adaptive regression splines in high-
dimensional inventory forecasting of water reservoir problem using stochastic dynamic programming, and
orthogonal array (OA) experimental designs. They also studied the implementation of artificial neural
networks (ANNs) as an alternative approach and found out that ANNs performed better with the OA-Latin
hypercube designs. Ambreen et al. (2019) studied to find out disparate stock administration systems of
assembling organizations in Pakistan to investigate the inventory management procedures functionally in the
country. The problems of stock administration determined in this study are poor production schedule, spillover,
stockout circumstance, high lead time and mishandling of records, it was determined that these parameters
have significant influence on the effectiveness of inventory management system. Otchere et al. (2016)
320
Abdulaziz S. ALKABAA, Osman TAYLAN
GU J Sci, Part A, 8(3): 318-326 (2021)
examined the existing inventory management practices of a chosen company in Ghana. The study used
interview administered questionnaire to collect data from staff of the company for keeping the stock always
available. The survey revealed that, the company faced with serious long lead time challenges in ordering parts
leading to cancellation of purchase orders, so losing customers.
2. MATERIAL AND METHOD
2.1. Orthogonal Array Design
There are several related meanings of orthogonal concept in statistical and mathematical sense. Statistically
speaking, two factors are orthogonal if they are statistically independent, a factor does not significantly affect
the other factor in an experimental design. Therefore, orthogonal array avoids factors from being too
complicated or redundant. The orthogonal design includes each of the level combinations (factor pairs) an
equal number of times. For illustration, consider three factors A, B, and C, each has two levels 1 and 2 as
shown in Table 1. It can be seen that the AB pairing (1, 1); (2, 1); (1, 2); and (2, 2) and other two pairings, AC
and BC, have only one occurrence or repeating, at each level combination.
Table 1. Orthogonal Design
Factor
Factor Levels
A
B
C
1
1
2
2
1
1
1
2
1
2
2
2
In contrast, the non-orthogonal design does not contain factor level pairing the same number of times. For
example, Table 2. displays AB level combinations; (1, 1); (2, 1); (1, 2); and (2,2). However, the other two
pairings AC and BC, have a duplicated and missed level combinations. For AC, (1, 2); (2, 1); (1, 2); (2, 1);
and the (2,2) combination is missing. For BC, (1, 2); (1, 1); (2, 1); (2, 1); and the (2, 2) combination is missing
and (1,2) is redundant.
Table 2. Non-Orthogonal Design
Factor
Factor Levels
A
B
C
1
1
2
2
1
1
1
2
1
2
2
1
Therefore, the features provided by orthogonal design help the analyst in constructing effective experiments
that can estimate each main effect and interaction independently. On the other hand, if the design is not
orthogonal, either by plan or by accidental loss of data, the results and explanations may be misguided.
The name of the orthogonal array is well known to statisticians in designing optimal experiments that require
complete or only a fraction (short duration) of all factorial combinations. An orthogonal array defines a
balanced layout, that is the factor levels are weighted alike. Then each factor can be evaluated separately from
all other factors, thus the influence of one factor does not impact the estimation of another factor.
321
Abdulaziz S. ALKABAA, Osman TAYLAN
GU J Sci, Part A, 8(3): 318-326 (2021)
Each column in an orthogonal array defines a specific factor that can have two or more levels. Each row
represents a run that indicates the factor level combinations for the run.
An orthogonal array OA is denoted by (N.f.l.t). The parameters N, f, l, and t represent the number of runs
(rows), the number of factors (columns), the number of levels, and strength of the design, respectively. For
illustration, Table 3 shows an OA (8.7.2.2) which includes only 8 runs which represents a fraction of the full
factorial design. On the other side, if a full factorial design were used, it would have 27 = 128 runs, which
consumes more costs, times, and efforts. It can be noticed that any 2 columns (factors) of the array of Table 3,
for example the first (A) and the third (C), each of the factor level combinations 11, 12, 21, and 22 occurs
exactly the same number of times (two in this case) and that represents the strength of the OA. In general,
orthogonal array designs emphasis mostly on major effects. However, some of the arrays allow a few selected
interactions to be studied. Orthogonal arrays are widely used in computer science, engineering, and business
Gopalakrishnan & Stinson (2008) and Garcia-Diaz & Phillips (1995).
Table 3. Orthogonal Array (8.7.2.2)
A
B
C
D
E
F
G
1
1
1
1
1
1
1
1
2
1
1
1
2
2
2
2
3
1
2
2
1
1
2
2
4
1
2
2
2
2
1
1
5
2
1
2
1
2
1
2
6
2
1
2
2
1
2
1
7
2
2
1
1
2
2
1
8
2
2
1
2
1
1
2
3. RESULTS AND DISCUSSION
The method starts by the analyst for creating a choice design that defines the main factors for constructing the
questionnaire. Each factor contains different levels or alternatives that properly define various choices
provided by the factors. A number of factors’ alternatives construct the options. A group of possible options
defines a choice set, and each set contains the choice of the questionaries. After, the options are constructed,
the total number of questions per respondent is defined based on different designs of orthogonal arrays (Sloane,
2007). For an illustration purpose, Figure 1 depicts the overall conceptual approach applied in this study.
Figure 1. Overall Conceptual Approach
322
Abdulaziz S. ALKABAA, Osman TAYLAN
GU J Sci, Part A, 8(3): 318-326 (2021)
Any manufacturing or distribution company’s success in today’s interrelated fast global supply chain logistic
environment is directly linked to the organization’s degree of inventory accuracy. However, the effort and
approach required to achieving inventory accuracy are often misunderstood and underestimated. Every year
the inventory committee people in the Electricity Company count and record physical items in warehouses and
comparing them with quantity in the system record; that is, the accuracy here is the percentage of matching
between the physical quantities and the recorded quantities in the inventory’s system, a traditional cycle
counting procedures are scheduled with no accuracy concern. Moreover, no suggested tools for accuracy
improvement are studied. The linked inventory systems become unreliable and uncertain, especially when that
happens in a critical and vital sector like Electricity Company warehouses. The problems associated with the
system unreliability affect the supply chains with many wrong purchasing orders created by the purchasing
department for the causes of shortage materials. As a result, excess materials and high costs for the inventory
will be.
One of various suggested solutions to the electricity company is to design a stated choice questionnaire for the
purpose of exploring and analyzing a lot of information to improve the inventory accuracy from the groups of
inventory engineers and workers with minimum number of questions. This application intends to study
different alternatives influence improving inventory accuracy. There are eight factors of interest: layout of
warehouse areas, material classification types, gate security, tracking material equipment, inventory system
structure, material documentation method, material handling type, and storage design. Table 4 lists the factors
and the associated levels to be considered in the study.
The proposed design for the survey is displayed in Figure 2 by the option or group of relevant factors that are
suggested to improve inventory accuracy.
As shown in Table 4, the orthogonal fractional factorial OA (16.8.2.3) was selected from the orthogonal array’s
library Sloane (2007) that allows sixteen questions, as it can be observed from Table 5 below of the stated
choice questionnaire, that each row of the design defines four options (A, B, C, D).
Table 4. Orthogonal array OA (16,8,2,3)
A
B
C
D
E
F
G
H
1
0
0
0
0
0
0
0
0
2
0
0
1
1
1
1
1
0
3
0
1
0
0
0
1
1
0
4
0
1
1
1
1
0
0
0
5
1
0
0
1
1
0
1
0
6
1
0
1
0
0
1
0
0
7
1
1
0
1
1
1
0
0
8
1
1
1
0
0
0
1
0
9
1
1
1
1
1
1
1
1
10
1
1
0
0
0
0
0
1
11
1
0
1
1
1
0
0
1
12
1
0
0
0
0
1
1
1
13
0
1
1
0
0
1
0
1
14
0
1
0
1
1
0
1
1
15
0
0
1
0
0
0
1
1
16
0
0
0
1
1
1
0
1
323
Abdulaziz S. ALKABAA, Osman TAYLAN
GU J Sci, Part A, 8(3): 318-326 (2021)
All questions asked to the same respondent are grouped as blocks. As presented in Table 5, the proposed stated
choice design suggests two blocks, as appears in Table 5 with a dashed line separated between them, with eight
questions each. Consequently, two respondents will be required to complete the questionnaire. This can be
completed, for example, by circling one of the letters A, B, C, D for each row shown on the last column of the
questionnaire of Table 5. The letters must be selected only once by each respondent. For example, respondent
number 1 may select C, A, D, and B for rows 1,2,5 and 8, respectively. Therefore, redundancy of options is
not allowed.
Figure 2. Factors and Levels that Affect in Improving the Inventory Accuracy
324
Abdulaziz S. ALKABAA, Osman TAYLAN
GU J Sci, Part A, 8(3): 318-326 (2021)
Table 5. Stated Choice Questionnaire for Improving Inventory Accuracy İn Electricity Company Warehouses
Set
Option A
Option B
Option C
Option D
Circle
Only
One
Option
Warehouse
Design
Security
Gate
Material
Receiving
Method
Documentation
Type
System
Structure
Tracking
Methodology
Storage
Design
Material
Handling
1
Internal
Design
Automated
Gate In
ABC
Method I
SAP
RFID
Shelf
Design
FIFO
A B C D
2
Internal
Design
Automated
Gate In
Priority
Method II
Oracle
Barcode
Internal
Layout
FIFO
A B C D
3
Internal
Design
Automated
Gate Out
ABC
Method I
SAP
RFID
Internal
Layout
FIFO
A B C D
4
Internal
Design
Automated
Gate Out
Priority
Method II
Oracle
Barcode
Shelf
Design
FIFO
A B C D
5
External
Design
Automated
Gate In
ABC
Method II
Oracle
Barcode
Internal
Layout
FIFO
A B C D
6
External
Design
Automated
Gate In
Priority
Method I
SAP
RFID
Shelf
Design
FIFO
A B C D
7
External
Design
Automated
Gate Out
ABC
Method II
Oracle
Barcode
Shelf
Design
FIFO
A B C D
8
External
Design
Automated
Gate Out
Priority
Method I
SAP
RFID
Internal
Layout
FIFO
A B C D
9
External
Design
Automated
Gate Out
Priority
Method II
Oracle
Barcode
Internal
Layout
LIFO
A B C D
10
External
Design
Automated
Gate Out
ABC
Method I
SAP
RFID
Shelf
Design
LIFO
A B C D
11
External
Design
Automated
Gate In
Priority
Method II
Oracle
Barcode
Shelf
Design
LIFO
A B C D
12
External
Design
Automated
Gate In
ABC
Method I
SAP
RFID
Internal
Layout
LIFO
A B C D
13
Internal
Design
Automated
Gate Out
Priority
Method I
SAP
RFID
Shelf
Design
LIFO
A B C D
14
Internal
Design
Automated
Gate Out
ABC
Method II
Oracle
Barcode
Internal
Layout
LIFO
A B C D
15
Internal
Design
Automated
Gate In
Priority
Method I
SAP
RFID
Internal
Layout
LIFO
A B C D
16
Internal
Design
Automated
Gate In
ABC
Method II
Oracle
Barcode
Shelf
Design
LIFO
A B C D
325
Abdulaziz S. ALKABAA, Osman TAYLAN
GU J Sci, Part A, 8(3): 318-326 (2021)
4. CONCLUSION
This paper presents the process of conducting orthogonal arrays to construct a stated-choice factorial survey.
Since the quality of the data gathering relies solely on the quality of questionnaire constructed and written
questions, applying the stated choice survey is intended to effectively gather the information that helps improve
the inventory accuracy in an electricity company warehouse. From a pool of several factors, eight factors with
two levels are selected to improve the accuracy problem. Four options that combine relevant factors in sixteen
sets of multiple choices are considered in the stated-choice factorial survey. The inventory's engineers and
workers will be targeted for data collection with more information with a minimum number of questions and
saving efforts, time, and costs of collecting the data. There are several issues needed to consider in designing
a stated-choice factorial survey. For instance, how far factor levels can deviate from the current knowledge of
the respondent and the order of presenting the questions.
ACKNOWLEDGMENTS
The author would also like to acknowledge that this work is not funded by any organization.
CONFLICTS OF INTEREST
The authors declare no conflict of interest.
REFERENCES
Abdi, H., & Williams, L. J. (2010). Principal component analysis. Wiley Interdisciplinary Reviews:
Computational Statistics, 2(4), 433-459. doi:10.1002/wics.101
Ambreen, N., Khan, S., & Fatima, U. (2019). A Survey on Contemporary Inventory Techniques and Problems
Faced By Manufacturing Organizations in Pakistan. In: Proceedings of 2019 IEEE 10th International
Conference on Mechanical and Intelligent Manufacturing Technologies (ICMIMT) (pp. 157-160). IEEE.
doi:10.1109/ICMIMT.2019.8712008
Bagozzi, R. P. (2011). Measurement and meaning in information systems and organizational research:
Methodological and philosophical foundations. MIS Quarterly, 35(2), 261-292. doi:10.2307/23044044
Boyes, A., Girgis, A., & Lecathelinais, C. (2009). Brief assessment of adult cancer patients' perceived needs:
development and validation of the 34‐item Supportive Care Needs Survey (SCNS‐SF34). Journal of
Evaluation in Clinical Practice, 15(4), 602-606. doi:10.1111/j.1365-2753.2008.01057.x
Brosnan, K., Grün, B., & Dolnicar, S. (2018). Identifying superfluous survey items. Journal of Retailing and
Consumer Services, 43, 39-45. doi:10.1016/j.jretconser.2018.02.007
Cervellera, C., Wen, A., & Chen, V. C. P. (2007). Neural network and regression spline value function
approximations for stochastic dynamic programming. Computers & Operations Research, 34(1), 70-90.
doi:10.1016/j.cor.2005.02.043
Dolnicar, S. (2013). Asking good survey questions. Journal of Travel Research, 52(5), 551-574.
doi:10.1177/0047287513479842
Fan, W., & Yan, Z. (2010). Factors affecting response rates of the web survey: A systematic review. Computers
in Human Behavior, 26(2), 132-139. doi:10.1016/j.chb.2009.10.015
Garcia-Diaz, A., & Phillips, D. T. (1995). Principles of experimental design and analysis. Springer.
Gopalakrishnan, K., & Stinson, D. R. (2006). Applications of orthogonal arrays to computer science. In: Proc.
of ICDM (pp. 149-164).
Ho, W., & Ma, X. (2018). The state-of-the-art integrations and applications of the analytic hierarchy process.
European Journal of Operational Research, 267(2), 399-414. doi:10.1016/j.ejor.2017.09.007
Hoerger, M. (2010). Participant dropout as a function of survey length in Internet-mediated university studies:
Implications for study design and voluntary participation in psychological research. Cyberpsychology,
Behavior, and Social Networking, 13(6), 697-700. doi:10.1089/cyber.2009.0445
326
Abdulaziz S. ALKABAA, Osman TAYLAN
GU J Sci, Part A, 8(3): 318-326 (2021)
Kolenikov, S. (2010). Resampling variance estimation for complex survey data. The Stata Journal, 10(2), 165-
199. doi:10.1177/1536867X1001000201
Liao, H., Xu, Z., & Zeng, X. J. (2015). Hesitant fuzzy linguistic VIKOR method and its application in
qualitative multiple criteria decision making. IEEE Transactions on Fuzzy Systems, 23(5), 1343-1355.
doi:10.1109/TFUZZ.2014.2360556
Meimand, S. E., Khalifah, Z., Zavadskas, E. K., Mardani, A., Najafipour, A. A., & Ahmad, U. N. U. (2017).
Residents’ attitude toward tourism development: A sociocultural perspective. Sustainability, 9(7), 1170.
doi:10.3390/su9071170
Mi, X., Li, J., Liao, H., Kazimieras Zavadskas, E., Al-Barakati, A., Barnawi, A., & Herrera-Viedma, E. (2019).
Hospitality brand management by a score-based q-rung ortho pair fuzzy VIKOR method integrated with the
best worst method. Economic Research, 32(1), 3266-3295. doi:10.1080/1331677X.2019.1658533
Otchere, A. F., Adzimah, E. D., & Aikens, I. (2016). Assessing the inventory management practices in a
selected company in Ghana. International Journal of Development and Sustainability, 5(3), 105-19, ISDS
Article ID: IJDS14122701
Sloane, N. J. A. (2007) "A Library of Orthogonal Arrays" http://neilsloane.com/oadir/
Verma, R., & Thompson, G. M. (1996). Basing service management on customer determinants: the importance
of hot pizza. The Cornell Hotel and Restaurant Administration Quarterly, 37(3), 18-23. doi:10.1016/0010-
8804(96)86809-1
Vetschera, R. (2017). Deriving rankings from incomplete preference information: A comparison of different
approaches. European Journal of Operational Research, 258(1), 244-253. doi:10.1016/j.ejor.2016.08.031
Yager, R. R. (2017). Generalized orthopair fuzzy sets. IEEE Transactions on Fuzzy Systems, 25(5), 1222-1230.
doi:10.1109/TFUZZ.2016.2604005