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

net/publication/242336470

Fuzzy logic and performance evaluation: Discussion and application

Article in International Journal of Productivity and Performance Management · March 2008

DOI: 10.1108/17410400810857248

CITATIONS READS
25 1,297

2 authors:

Hooshang Beheshti James G. Lollar

Radford University Radford University
39 PUBLICATIONS 1,349 CITATIONS 15 PUBLICATIONS 278 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Hooshang Beheshti on 04 February 2015.

The user has requested enhancement of the downloaded file.

 The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1741-0401.htm

Fuzzy logic and

Fuzzy logic and performance performance
evaluation: discussion and evaluation
application
237
Hooshang M. Beheshti
Department of Management, College of Business and Economics, Received March 2007
Radford University, Radford, Virginia, USA, and Revised July 2007
Accepted July 2007
James G. Lollar
Department of Marketing, College of Business and Economics,
Radford University, Radford, Virginia, USA

Abstract
Purpose – This paper seeks to describe the development of the fuzzy logic model approach to
decision making and its value for managers by illustrating its application to employee performance
appraisals.
Design/methodology/approach – An extensive literature review provided the framework for the
model development in this research. Performance evaluations represent a critically important decision
that often involves subjective information. Models and heuristic techniques that focus on the use of
different types of information are available; however, with few exceptions, the models are not robust
enough to be applied in a practical, managerially useful manner. Fuzzy logic models provide a
reasonable solution to these common decision situations.
Findings – Fuzzy logic can be a powerful tool for managers to use instead of a traditional
mathematical model when evaluating the performance of personnel or teams. The flexibility of the
model allows the decision maker to introduce vagueness, uncertainty, and subjectivity into the
evaluation system.
Research limitations/implications – This research calls attention to an alternative method of the
performance evaluation system as opposed to the traditional quantitative methods. Future research in
this area is needed to develop a method for relating membership values to linguistic variables in
performance evaluation, as well as testing the sensitivity of membership values and their impact on
the outcome.
Originality/value – This paper provides a simple-to-use fuzzy logic model for establishing a more
meaningful evaluation system.
Keywords Performance appraisal, Fuzzy logic, Decision making, Linguistics
Paper type Research paper

Managers encounter many decisions that require the simultaneous use of different
types of data in their decision-making process. A critical decision area for managers is
the performance evaluation of personnel, whether individually or as a member of a
team. Performance evaluations require and often involve disparate types of
information that are vague, incomplete, objective, and subjective. The importance of International Journal of Productivity
performance evaluation of personnel and its relationship to the firm’s performance is and Performance Management
Vol. 57 No. 3, 2008
well documented in the literature (Lowe, 1986; Katzenbach and Smith, 1993; Kilduff pp. 237-246
et al., 2000; Higgs, 2005). A variety of appraisal techniques are available to measure q Emerald Group Publishing Limited
1741-0401
performance (Chang and Hahn, 2006, Jiang et al., 2001, Armentrout, 1986; Arvey and DOI 10.1108/17410400810857248
IJPPM Murphy, 1998; Sanchez and De La Torre, 1996; Stronge, 1991). In creating and
57,3 implementing an appraisal system, management must determine what the
performance appraisal system will be used for, and then decide on the process to
implement the system. The methods chosen and the instruments used to implement
these methods are crucial in determining whether the organization manages its
performance successfully.
238 These appraisal methods are based on quantitative techniques which provide a
precise output defining the employees’ performance. However, much of the information
related to performance appraisal is not quantifiable and precise with crisp boundaries.
Rather, this information is presented in expressions or words in natural language and
without precision.
A useful approach for examining many real-world problems is fuzzy approximate
reasoning or fuzzy logic. This technique is based on the fuzzy set theory (Zadeh, 1965)
that allows the elements of a set to have varying degrees of membership, from a
non-membership grade of 0 to a full membership of 100 per cent or grade 1. This
smooth gradation of values is what makes fuzzy logic match well with the vagueness
and uncertainty typical of many real world problems. In fuzzy logic, a statement such
as “Sam is old” can be 60 per cent true and 40 per cent false; i.e. Sam has a 0.60 degree
of membership in a set whose elements are considered to be old and a 0.40 degree of
membership in a set of young people. A person can be a member of both sets at the
same time. Although the sum of the degree of membership is 100 per cent in our
example, fuzzy logic does not require it.
For a person to be a member of both old and young sets violates the basic law of
non-contradiction of conventional logic in which an element either belongs (grade: 1) or
does not belong (grade: 0). Fuzzy logic allows this violation by generalizing properties
of traditional sets into fuzzy counterparts, thus providing a greater expressive power
than traditional logic.

Background
Fuzzy set theory (Zadeh, 1965) was developed to address contexts in which decision
makers need to accurately analyze and process information that is imprecise in nature.
Fuzzy sets provide a conceptual framework, as well as an analytical tool to solve real
world problems where there is a lack of specific facts and precision (Baldwin, 1996; Klir
and Yuan, 1995; Kosko, 1991). Since its introduction, fuzzy set and fuzzy logic have
been applied to a wide range of scientific and engineering problems. For example,
fuzzy models have been used to develop microchips that run fuzzy algorithms.
Presently, a variety of consumer products such as television sets, automobiles,
camcorders and cameras use fuzzy logic chips; fuzzy industrial controllers for
smoother operation of trains, automobiles, and elevators are in use today; military
hardware and operations, as well as robotics are using fuzzy logic for better efficiency
(Ross, 2004; Pires and Hoshiau, 2002; Rogers and Hoshiau, 1990; Kosko, 1991).
However, the application of fuzzy set theory and logic to management decisions has
been generally lacking despite its potential value in many common situations (Dorsey
and Coovert, 2003; Alliger et al., 1993). The purpose of this paper is to develop a
framework for the application of fuzzy logic to employee performance appraisal.
Although our example is focused on employee performance appraisal, the model can be
used to evaluate the performance of teams and business units in an organization.
 Extensive research during the past 50 years has focused on the value and inherent Fuzzy logic and
problems of employee performance evaluations (Stajkovic and Luthans, 2003). performance
Numerous models and frameworks have been proposed and tested to address the
complexities of dynamic work environments. With regard to the effect of performance evaluation
evaluations on work outcomes, most empirical research have used quantitative,
objective data that is collected systematically as part of an organization’s performance
appraisal process. 239
Fuzzy sets structure and analysis
The most fundamental form of a fuzzy set A in a universe X is:
A ¼ {x; mA ðxÞjx1X}
where mA(x) represents the grade of membership or compatibility function of element x
of X in fuzzy set A. Element x may show a full membership in A (i.e. mA(x) ¼ 1), as well
as partial membership (0 , mA(x) , 1) or non-membership (mA(x) ¼ 0). For example,
the fuzzy linguistic variable leadership can be characterized by terms: very strong,
strong, average, weak, poor, and very poor. Each term is called a linguistic modifier.
Therefore, a fuzzy set is formed when a linguistic variable is combined with a linguistic
modifier (i.e. strong leadership). In our example, each linguistic modifier is linked to a
numerical value on a scale of 1 to 7 that represents the level of leadership. Thus, the
leadership set A and its modifiers can be represented by a fuzzy set as:
A ¼ {1:0j0; 2:0j0:10; 3:0j0:30; 4:0j0:40; 5:0j0:60; 6:0j0:80; 7:0j1:0}:
In this fuzzy set, each element represents a corresponding value in the universe of
discourse and a degree of membership. That is, 7 has a full membership grade of 1
corresponding to very strong leadership, and 1 with a non-membership grade of 0
indicating no leadership as well as 5 with a partial membership grade of 0.60
representing average leadership. Fuzzy set A could also be shown as depicted in
Figure 1. The graph in Figure 1 shows a gradual transition of leadership from very
poor to very strong, as opposed to sudden jumps from one value to another.
Another example of constructing fuzzy sets for linguistic variables is presented in
Figure 2, where three fuzzy sets are used to characterize a person’s performance. The
fuzzy linguistic variable performance can be defined by terms or linguistic modifiers

Figure 1.
Fuzzy set depiction of
leadership
IJPPM
57,3

240
Figure 2.
Fuzzy set structure of
performance

poor, average, outstanding, with the membership value from 0 to 1 describing the level
of performance on a scale from 0 to 5.
In Figure 2, if the norm for performance is average, number 3 represents the highest
level of the term average with a membership grade of 1 and number 4 defines average
with a grade of 0.10 or outstanding with a grade of 0.60. Therefore, number 4 describes
the performance of a person whose performance is 60 per cent outstanding and 10 per
cent average. Figure 2 represents three fuzzy sets:
Poor performance ¼ {0j1:0; 0:50j0:80; 1:0j0:50; 1:50j0:40; 2:0j0:30;
3:0j0:0; 3:50j0:0; 4:0j0:0; 4:50j0:0; 5:0j0:0}

Average performance ¼ {0j0:0; 0:50j0:0; 1:0j0:0; 1:50j0:0; 2:0j0:10; 2:50j0:40; 3:0j1:0;

3:50j0:50; 4:0j0:10; 4:50j0:0; 5:0j0:0}

Outstanding performance ¼ {0j0:0; 1:0j0:0; 1:50j0:0; 2:0j0:0; 2:50j0:0; 3:0j0:0; 3:50j0:20;

4:0j0:60; 4:50j0:80; 5:0j1:0}:

These sets represent the decision maker’s intuitive understanding of the linguistic
variable performance and its modifiers: poor, average, and outstanding.

Fuzzy set operations

Fuzzy sets can be manipulated by one of the four standard fuzzy set operations: union,
intersection, complementation, and implication operations (Schmucker, 1984; Mendel,
2001). To illustrate, assume A and B are fuzzy sets with membership functions mA
(x) ¼ {0,3,5,7,8}, with unit interval x ! [0,10], mB (y) ¼ {1, 2, 4, 6, 9}, and y ! [0,10]
respectively. The union of A and B is a fuzzy set C ¼ A<B where:
mA _B ðzÞ ¼ mA ðxÞ _ mB ðyÞ:
A union operation is identical to a logical OR operation and a fuzzy set union is
performed by applying the max function to the elements of two sets; that is, taking the
higher value of the first, second, . . . element in each set in order to form the union of the Fuzzy logic and
two sets, thus:
performance
mA _B ðzÞ ¼ {1; 3; 5; 7; 9}: evaluation
A logical AND can be used to determine a fuzzy set D ¼ A>B with:
mA ^B ðwÞ ¼ mA ðxÞ ^ mB ðyÞ 241
fuzzy set intersections are done by applying the min function; therefore:
mA ^B ðwÞ ¼ {0; 2; 4; 6; 8}:
The complement of a set is computed by subtracting each element of the set from its
maximum possible value, in our example 10. Thus:
mA
ðxÞ ¼ 10 2 mA ðxÞ ¼ {10; 7; 5; 3; 2}
and
mB
ðyÞ ¼ 10 2 mB ðyÞ ¼ {9; 8; 6; 4; 1}:
The implication function is used to decide if A is true, to what extent does that imply
that B is true? The implication operation is done by computing A
< B, known as
Kleene-Dienes implication (Whalen and Schott, 1992), where:
mA
_B ðuÞ ¼ mA
ðxÞ _ mB ðyÞ

mA
_B ðuÞ ¼ {10; 7; 5; 6; 9}:

It should be noted that fuzzy set operations are not limited to those used here; for other
fuzzy operations, see (Mendel, 2001; Klir and Yuan, 1995; Kosko, 1991).

Fuzzy performance appraisal

Performance evaluations of employees for rewards such as salary increases and
promotions are not easy; many criteria and standards that are involved should be
considered. However, for illustration purposes and to keep matters relatively simple to
follow, the example used in this section includes seven employees and six categories of
performance evaluations for each employee. These categories, as shown in Table I,
form a fuzzy set C in a universe U with the unit interval [0,1], where:

Category Relative importance

Task accomplishment (TA) 1.00

Decision making (DM) 0.60
Committee work (CW) 0.40
Human relations (HR) 0.80
Communications (CO) 0.90 Table I.
Teamwork (TW) 0.70 Category importance
IJPPM C ¼ {ujmC ðuÞ; u1U }
57,3
mC ðuÞ ¼ {1:0; 0:60; 0:40; 0:80; 0:90; 0:70}:
Each element of the set is given a score between 0 and 1; a high score signifies the
relative importance of that category (fuzzy element) to the decision maker. Equal
242 membership means equal importance.
For each of the six categories, a qualitative judgment is used to determine the degree
of employee performance for that category. These qualitative judgments could be: “not
acceptable”, “poor”, “below average”, “slightly below average”, “average”, “slightly
above average”, “above average”, and “outstanding”, thus forming a fuzzy set P in
universe V with unit interval [0, 1] and a fuzzy membership function:
P ¼ {vjmP ðvÞ; v1V }

mP ðvÞ ¼ {0:10; 0:20; 0:30; 0:40; 0:60; 0:70; 0:80; 1:0}

as shown in Table II.
To illustrate how the manipulation of fuzzy sets can result in a decision making
system for performance evaluation, several steps must be taken. The first step is to
assess the performance of each employee by each category that is based on the fuzzy
opinion of the evaluator, as depicted in Table III.
The next step is to use Tables II and III to generate Table IV representing
membership grades of employees for all categories.
Table IV contains seven fuzzy sets, P1,. . .,P7, with membership functions
mP1(v),. . .,mP7(v). For example, the fuzzy set and membership function for employee 1 is:

Category Relative importance

Not acceptable (NA) 0.10

Poor (P) 0.20
Below average (BA) 0.30
Slightly below average (SBA) 0.40
Average (Ave) 0.60
Table II. Slightly above average (SAA) 0.80
Membership value of Above average (AA) 0.90
performance Outstanding (O) 1.00

Employee (EM) performance rating

Performance category EM1 EM2 EM3 EM4 EM5 EM6 EM7

TA BA Ave AA SAA Ave SAA AA

DM P Ave SAA Ave AA Ave AA
CW SAA AA SAA AA SBA SBA AA
Table III. HR AA Ave AA SBA AA Ave O
Employee performance CO Ave SAA SAA Ave SBA AA AA
by category TW Ave AA AA P SAA O O
 P 1 ¼ {1:0j0:30; 2:0j0:10; 3:0j0:70; 4:0j0:80; 5:0j0:60; 6:0j0:60} Fuzzy logic and
performance
mP1 ðvÞ ¼ {0:30; 0:10; 0:70; 0:80; 0:60; 0:60}: evaluation
The principal step in the decision making process is to establish a fuzzy implication
relation between a specific category and each employee’s performance for that category.
That is to say, given the relative importance of a category, does that imply a good 243
performance by the employee for that category? Assuming that the importance assigned to
each category is the maximum value for that category, the implication relationship is
established by taking the complement of the category importance.
This complementation creates a minimum performance value assigned to all
employees given the category. The max function is applied to each employee’s
performance set, i.e. mP1(v),. . .,mP7(v) and the complement of set mC(u). Thus:
mC
_P1 ðrÞ ¼ mC
ðuÞ _ mP1 ðvÞ

mC
_P1 ðrÞ ¼ {0; 0:40; 0:60; 0:20; 0:10; 0:30} _ {0:30; 0:10; 0:70; 0:80; 0:60; 0:60}
¼ {0:30; 0:40; 0:70; 0:80; 0:60; 0:60}:

The final step is to combine various performances of the employee across all categories
in order to obtain an overall evaluation. This is done by applying the min function to
the set derived from the previous step. Table V shows the overall rating of the
employees once the process is completed.
It is not surprising that employee 7 has been ranked top performer in this example
since the proposed system favors the employee with a high rating in the most
important category. Thus, the higher the relative importance the category is, the more
influence that category has in the final output.

Category EM1 EM2 EM3 EM4 EM5 EM6 EM7

1 (TA) 0.30 0.40 0.80 0.70 0.60 0.70 0.80

2 (DM) 0.10 0.60 0.70 0.60 0.80 0.60 0.80
3 (CW) 0.70 0.80 0.40 0.80 0.40 0.40 0.80
4 (HR) 0.80 0.60 0.80 0.40 0.80 0.60 1.00 Table IV.
5 (CO) 0.60 0.70 0.70 0.60 0.40 0.80 0.80 Membership grades of
6 (TW) 0.60 0.80 0.80 0.20 0.70 1.00 1.00 performance by category

Employee Score

1 0.30
2 0.60
3 0.70
4 0.30
5 0.40
6 0.60 Table V.
7 0.80 Employees overall rating
IJPPM Conclusions
57,3 Traditionally, performance appraisal methods were broken down into two categories
based on the standards for success chosen, comparative methods ranking, or forced
distribution, to rate the overall performance of one employee or a team against that of
other employees or teams. Absolute standards methods, such as rating scales or
management by objective (MBO), rate the employee or team against some objective or
244 management goal(s). In both cases, managers often use a quantitative tool to perform this
task (Miller and Thor, 2006; Capaldo and Zollo, 2001; Smith et al., 1996; Stronge, 1991).
The use of quantitative techniques provides a tool for decision makers to arrive at a
solution that is easily explained and understood. However, many factors used in the
process are subjective and difficult to quantify. Fuzzy logic enables the reviewer or the
decision maker to incorporate information in the evaluation system which is vague and
subjective.
There are several advantages in using the model presented in this paper as opposed
to a traditional and complex quantitative technique. The mathematics is extremely
simple and can be easily computerized using a spreadsheet or a database system. It is
also extremely flexible, allowing the decision maker to use a broad range of linguistic
variables and modifiers for finer discrimination or to make changes to membership
values and/or performance categories. Finally, it is an ideal system when the decision
maker is faced with a series of sub-decisions where available data is based on
vagueness, uncertainty, and opinion. These sub-decisions are then combined into an
overall system for performance evaluation.

References
Alliger, G.M., Feinzig, S.L. and Janak, E.A. (1993), “Fuzzy sets and personnel selection: discussion
and an application”, Journal of Occupational and Organizational Psychology, Vol. 66 No. 2,
pp. 163-9.
Armentrout, D.R. (1986), “Engineering productivity management and performance
measurement”, Journal of Management in Engineering, Vol. 2 No. 3, pp. 141-7.
Arvey, R.D. and Murphy, K.R. (1998), “Performance evaluation in work settings”, Annual Review
of Psychology, Vol. 49, pp. 141-68.
Baldwin, J.F. (1996), Fuzzy Logic, John Wiley & Sons, New York, NY.
Capaldo, G. and Zollo, G. (2001), “Applying fuzzy logic to personnel assessment: a case study”,
The International Journal of Management Science, Vol. 29 No. 6, pp. 585-97.
Chang, E. and Hahn, J. (2006), “Does pay-for-performance enhance perceived distributive justice
for collectivistic employees?”, Personnel Review, Vol. 35 No. 4, pp. 397-412.
Dorsey, D.W. and Coovert, M.D. (2003), “Mathematical modeling of decision making: a soft and
fuzzy approach to capturing hard decision”, Human Factors, Vol. 45 No. 1, pp. 117-35.
Higgs, M. (2005), “Influence of team composition and task complexity on team performance”,
Team Performance Management, Vol. 11 Nos 7/8, pp. 227-50.
Jiang, J.J., Sobol, M.G. and Klein, G. (2001), “A new view of IS personnel performance evaluation”,
Communications of the ACM, Vol. 44 No. 6, pp. 95-102.
Katzenbach, J.R. and Smith, D.K. (1993), The Wisdom of Teams: Creating the High Performance
Organization, Harvard Business School Press, Boston, MA.
Kilduff, M., Angelmar, R. and Mehra, A. (2000), “Top management team diversity and firm Fuzzy logic and
performance: examining the role of cognitions”, Organization Science, Vol. 11 No. 1,
pp. 21-34. performance
Klir, G.J. and Yuan, B. (1995), Fuzzy Sets and Fuzzy Logic, Prentice-Hall, Upper Saddle River, NJ. evaluation
Kosko, B. (1991), Neural Networks and Fuzzy Systems, Prentice-Hall, Upper Saddle River, NJ.
Lowe, T.R. (1986), “Eight ways to ruin a performance review”, Personnel Journal, Vol. 65 No. 1,
pp. 60-2. 245
Mendel, J.M. (2001), Uncertain Rule-Based Fuzzy Logic Systems: Introduction and New Directions,
Prentice-Hall, Upper Saddle River, NJ.
Miller, C.E. and Thor, C.L. (2006), “How accurate are your performance appraisals?”, Public
Personnel Management, Vol. 35 No. 2, pp. 153-62.
Pires, G. and Hoshiau, Y. (2002), “A wheelchair steered through voice commands and assisted by
a reactive fuzzy logic controller”, Journal of Intelligent and Robotics Systems, Vol. 34 No. 3,
pp. 301-14.
Rogers, M. and Hoshiau, Y. (1990), “The future looks fuzzy”, Newsweek, 28 May, pp. 46-7.
Ross, T.J. (2004), Fuzzy Logic with Engineering Applications, John Wiley, New York, NY.
Sanchez, J.I. and De La Torre, P. (1996), “A second look at the relationship between rating and
behavioral accuracy in performance appraisal”, Journal of Applied Psychology, Vol. 81
No. 1, pp. 3-10.
Schmucker, K.J. (1984), Fuzzy Sets, Natural Language Computations and Risk Analysis, Computer
Science Press, Rockville, MD.
Smith, B.N., Hornsby, J.S. and Shirmeyer, R. (1996), “Current trends in performance appraisal:
an examination of managerial practice”, SAM Advanced Management Journal, Vol. 61
No. 3, pp. 10-15.
Stajkovic, A. and Luthans, F. (2003), “Behavioral management and task performance in
organizations: conceptual background, meta-analysis, and test of alternative models”,
Personnel Psychology, Vol. 56 No. 1, pp. 155-95.
Stronge, J.H. (1991), “The dynamics of effective performance evaluation systems in education:
conceptual, human relations, and technical domains”, Journal of Personnel Evaluation in
Education, Vol. 5 No. 1, pp. 77-83.
Whalen, T. and Schott, B. (1992), “Presumption, prejudice, and regularity in fuzzy material
implication”, in Zadeh, L.A. and Kacprzyk, J. (Eds), Fuzzy Logic for the Management of
Uncertainty, John Wiley & Sons, New York, NY.
Zadeh, L.A. (1965), “Fuzzy sets”, Information and Control, Vol. 8 No. 3, pp. 338-53.

About the authors

Hooshang M. Beheshti is professor and chair of the Department of Management at Radford
University, Radford Virginia. He has over 30 years of experience as an educator and consultant.
He has co-authored two textbooks, one in management science and one in information systems
and has published extensively on a variety of management topics. His articles have appeared in
the Journal of Computer Information Systems, Journal of Business and Entrepreneurship, Journal
of International Information Management, Journal of Total Quality Management and Business
Excellence, International Journal of Service Business, Competitiveness Review, and Industrial
Management and Data Systems among others. He is one of the co-founders of the International
Academy of Business Disciplines (IABD) as well as one of its past presidents. Hooshang
M. Beheshti is the corresponding author and can be contacted at: [email protected]
IJPPM James G. Lollar is currently Associate Professor and acting Chairman of Marketing at
Radford University. Dr Lollar earned his PhD from the University of Alabama. Previous to
57,3 entering academe, he worked for Exxon Corporation in numerous sales and marketing positions,
and has served as a consultant to more than 100 regional, national, and international
organizations. Dr Lollar’s previous publications have appeared in the Journal of Marketing
Research, Journal of Advertising Research, Journal of Personal Selling and Sales Management,
Journal of Services Marketing, Journal of Marketing for Higher Education, and others. His
246 primary research interest is the investigation of buyer-seller relationships, the
internationalization process of firms, and information processing effects on decisions.

To purchase reprints of this article please e-mail: [email protected]

Or visit our web site for further details: www.emeraldinsight.com/reprints

View publication stats