theory for performance indicators
TRANSCRIPT
-
8/13/2019 Theory for performance Indicators
1/137
POLITECNICO DI TORINODept. of Production Systems and Business Economics
Performance indicators
by Domenico Maisano
Course of
Quality Engineering
-
8/13/2019 Theory for performance Indicators
2/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
2
Quality
(of a product or service) refers to the degree
to which a set of inherent characteristics fulfills (stated
or implied) requirements/needs (ISO-9000, 2000).In other words, Quality is the ability to fulfill different
types of requirements (e.g., productive, economical,social) with appropriate and measurable actions.
Quality and Indicators
-
8/13/2019 Theory for performance Indicators
3/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
3
to achieve Quality is therefore necessary:
1) To identify/observe these needs;2) To fulfill them using the appropriate resources.
Indicators are essential tools for observing
the
evolution of the process and its context.
Quality and Indicators
-
8/13/2019 Theory for performance Indicators
4/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
4
A Quality Management Systemis a set of tools for
driving and controlling an organization, considering all
different Quality aspects (ISO-9000, 2000):
human resources; know-how and technology; working practices, methodologies and procedures.
Quality guidelines
-
8/13/2019 Theory for performance Indicators
5/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
5
Typical requirements
are production, cost, time, ROI,
stakeholders exigencies or expectations.
Typical monitoring
activities are:
performance evaluation of the whole firm aspects(processes, suppliers, employees, Customer Satisfaction); market analysis (shares, development opportunities); productivity and competitors analysis; decisions about product innovation or new servicesprovided.
Quality guidelines
-
8/13/2019 Theory for performance Indicators
6/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
6
Modeling
a system means describing/representing
it,
considering the targets which should be met.A proper performance measurement system should be
arranged to verify if responses are consistent withrequirements.
A system of indicators should become an information system
for estimating the level of achievement of quality targets (UNI-11097, 2003).
Observation, representation, modelling
-
8/13/2019 Theory for performance Indicators
7/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
7
The Importance of Indicators
Indicators are supposed to be a representationmodel of a generic system/process.As a consequence they are the basis for
evaluations, judgments or decisions.
-
8/13/2019 Theory for performance Indicators
8/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
8
The importance of indicators
ISO 9004/2: 2009 standard explains how organizations can use a qualitymanagement approach to achieve sustained success.
8.3 MEASURING
ORGANIZATIONAL PERFORMANCE
8.3.1 GENERAL APPROACH TO ASSESSING PROGRESS
Assess your progress by measuring
your actual achievements against the results you plan to achieve.
Use a measurement
and analysis process to monitor your actual achievements against the results you
plan to achieve.
Select appropriate and practical methods for collecting information and monitoring key performanceindicators.
8.3.2 DEFINING KEY PERFORMANCE INDICATORSIdentify factors that are critical to your organizations success.
Select your organizations key performance indicators
(KPIs).
Implement KPIs
throughout your organization.
Use your KPIs to measure performance.
-
8/13/2019 Theory for performance Indicators
9/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
9
They may influence
and modify actions and decisions.
If a firm measures indicators a, b and c, neglecting x,y and z, then managers will pay more attention to thefirst ones. The firm gains core strengths in producing a,b and c.Firms become what they measure!(Hauser and Katz, 1998).
If maximizing a, b and c leads to long-term profit, theindicators are effective. If a, b and c lead tocounterproductive decisions and actions, then indicatorshave failed.
But even worse! Once the enterprise is committed to theseindicators, indicators provide tremendous inertia.
Indicators also have a deep normative effect
-
8/13/2019 Theory for performance Indicators
10/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
10
Measurements, Estimators,
Performance Indicators, Metrics...
Indicators are used within a wide range offields:
operations management (logistics, production,quality management, marketing, etc...);
Economy and finance;Social sciences;
Sport races (gymnastics, diving, figure skating,
etc...).
-
8/13/2019 Theory for performance Indicators
11/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
11
The customers perception on a new product;
The personal name of an individual;
The cycle time of a manufactured product.
Social status indicators (GDP):
- e.g. European countries with debt-to-GDPratio lower than 3% can adopt Euro currency;
- country inflation is running at 2.7 %;the air quality index value is 6 and so on
Some Examples
-
8/13/2019 Theory for performance Indicators
12/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
12
Open questions
Dealing with indicators, many questions arise:
How many indicators shall we use?
Is there an optimal set?
Is this set unique?
If not, what is the best one (if it exists)?
Can all these indicators be aggregated in a unique one?
What are the properties of indicators?
Are indicators the same as measurements?
Are there operative methods for defining and testing theindicators of a generic process?
-
8/13/2019 Theory for performance Indicators
13/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
13
Measurements (according to tradition)
According to physicists:
Measurement: is the process by which one canconvert physical parameters to meaningful numbers.
Instrument: may be defined as a device fordetermining the value or magnitude of a quantity orvariable.
The standard measure of each kind of physicalquantity is the unit; the number of times the unitoccurs in any given amount of the same quantity is the
number of measure. Without the unit, the number ofmeasure has no physical meaning.
-
8/13/2019 Theory for performance Indicators
14/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
14
Measurements (according to tradition)
The basic units are called fundamentals, while all
the others which can be expressed in terms offundamental units are called derived units, andformed by multiplying or dividingfundamental units.
E.g., the primary fundamental units which mostcommonly used are length, mass, and time, whilemeasurement of certain physical quantities in thermal,electrical, and illumination disciplines are alsorepresented by fundamental units.
Every derived unit originates from some physical law
defining that unit.
-
8/13/2019 Theory for performance Indicators
15/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
15SI:
InternationalS
ystemo
fUnits
(Sy
stmei
nternationald'units)
-
8/13/2019 Theory for performance Indicators
16/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
16
What about the measurement of non-physicalquantities?
e.g., raw scores of intelligence; grades of leather;
guidability of a car.WARNING! In many cases measurement
procedures are not as much well-defined andcannot be objective.
Any other types of Measurements?
-
8/13/2019 Theory for performance Indicators
17/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
17
A measurementis the assignment of numbers to properties of objectsor events in the real world, by means of an objectiveand empirical
operation, to describe them.objective
independence on the subjects (judgments may be
repeated by different subjects, obtaining the same result);
empirical it comes from reality (i.e., its not pure theory);objects/events and the corresponding properties/relations areobservable.
For example, an evaluation is not a measurement because of the lackof objectivity (it may depends on subjective perceptions).-
evaluations on the importance of customer needs according to
customers (within HoQ)
-
evaluations about the teaching activity of your teacher(s)
Some definitions according to S.S. Stevens
-
8/13/2019 Theory for performance Indicators
18/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
18
The Theory of Scales of Measurement by S.S. Stevens
Note: The columns listing the basic operations and the permissible statistics arecumulative: to an operation/statistic listed must be added all thoseoperations/statistics preceding it. Conversely, the column about
the
mathematical transformations which leave the scale-form invariant is inverselycumulative: each transformation in the column is contained in thetransformation immediately above it.
Scale
Type
Basic Empirical
Operations (relations
among objects)
Permissible scale-
transformations
Permissible Statistics Examples
Nominal Equivalence (equality) Permutation (one-to-
one substitution)
Mode, chi square Eye colour, place of
birth, etc
Ordinal Order (greater or less) Monotonic
increasing function
Median, percentiles Surface hardness,
military rank, etc
Interval Distance (addition orsubtraction)
Linear function(x=ax+b)
Mean, standard deviation,correlation, regression,
analysis of variance
Temperature in C,serial numbers, etc
Ratio Ratio (multiplication or
division)
Similarity (x=ax) geometric mean, harmonic
mean, coefficient of
variation, logarithms
Temperature in K,
weight, age, number
of children, etc
Tab. 1 Classification scheme of measurements/indicators depending on their scale types
CARDINAL
CATEGORICAL
No Oil In Rivers
-
8/13/2019 Theory for performance Indicators
19/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
19
Nominal ScaleThe numbers/symbols serve only as labels
or tags for identifying and
classifying objects.
e.g., classes concerning marital status: (A)-Married, (B)-Divorced,
(C)-Never married, (D)-Widowed.When used for identification, there is a strict one-to-one correspondence
between the numbers and the objects.e.g., numbering
of football players for the identification of the individuals.
Since the purpose is just as well served when any two designatingnumerals are interchanged, this scale form remains invariant under thegeneral substitution or permutation group.
Only a limited number of statistics, all of which are based on frequencycounts, are permissible, e.g., percentages, and mode.
-
8/13/2019 Theory for performance Indicators
20/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
20
Ordinal Scale
A ranking scale in which numbers are assigned to objects to indicatethe relative extent
to which the objects possess some characteristic.
Can determine whether an object has more or less of a characteristicthan some other object, but not how much more or less
e.g. T-shirt sizes: XS, S, M, L, XL and so on.
Any series of numbers can be assigned that preserves the ordered relationships between the objects (transitivityproperty of inequalitieshas to be satisfied too). Strictly (monotone) increasing transformationsof scale are permissible.
In addition to the counting operation allowable for nominal scale data,ordinal scales permit the use of statistics based on centiles, e.g.,percentile, quartile, median.
-
8/13/2019 Theory for performance Indicators
21/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
21
Interval ScaleNumerically equal distances on the scale represent equalvalues in the characteristic being measured.It permits comparison of the differences between objects.The location of the zero point is not fixed. Both the zeropoint and the units of measurement are arbitrary (e.g.altitude above sea level).
Any positive linear transformation of the form(x) = a + bx (with b>0) will preserve the properties ofthe scale.It is not meaningful to take ratios of scale values.
Statistical techniques that may be used include all of thosethat can be applied to nominal and ordinal data, and inaddition the arithmetic mean, standard deviation, andother statistics commonly used.
-
8/13/2019 Theory for performance Indicators
22/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
22
Ratio ScalePossesses all the properties of the nominal, ordinal, and intervalscales.
It has an absolute zero point (generally corresponding to theabsence of manifestation of the characteristic being measured).It is meaningful to compute ratios of scale values.
Only proportionate transformations of the form (x) = bx, where
b is a positive constant, are allowed.All statistical techniques can be applied to ratio data.
A scale with no arbitrary choice of a unit
or zero
to be made is
called absolute scale(e.g., the number of members of a givencollection of objects is determined uniquely, or the probability
of a
certain event [0,1]). The absolute scaleis a special variantofratio scales.
invariant just under the (trivial) identity transformation (x)=x.
-
8/13/2019 Theory for performance Indicators
23/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
23
The Mohs
scale of mineral hardness characterizes the scratch
resistance of various minerals through the ability of a harder materialto scratch a softer material.
The Mohs
scale is based on 10 standard minerals:
Link between observed relations and numerical relations
1. Talc
2. Gypsum3. Calcite
4. Fluorite
5. Apatite
6. Feldspar
7. Quartz8. Topaz
9. Corundum
10. Diamond
The Mohs
scale is strictly an ordinalscale (no interval property) and
uses half-numbers for in-between hardnesses. For instance, dolomite,which scratches calcite but not fluorite, has a Mohs
hardness of 3.5.
The ordinal relation between materials in terms of hardness is reflectedby the ordinal relation among the corresponding numbers(ISOMORPHISM = same structure).
-
8/13/2019 Theory for performance Indicators
24/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
241 kg 2 kg 5 kg 8 kg
A BC D
Weights, lengths and resistances are fundamental magnitudes defined onratio scales.
They satisfy the criterion of additivity, in fact they can be added in thephysical sense.
The corresponding relations (e.g., larger than, commutative and
associative property, transitive property of (in)equalities, etc) are reflectedby the relations among numbers.
We can make statements based on numbers, which reflect the relationsamong real objects
Link between observed relations and numerical relations
For example:
1)
C is heavier that B, which is heavierthan A
2)
A, B and C added together weigh asmuch as D
-
8/13/2019 Theory for performance Indicators
25/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
25
A
transformation of scale
is admissible
when it preserves the basic
relations among objects (represented on the scale).
-
see the 2nd
column of the previous table.
It can be demonstrated mathematically that every scale type admitsonly some specific transformations of scale (see Roberts).
Other definitions
-
8/13/2019 Theory for performance Indicators
26/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
26
A
statement/assertion using scales is called meaningful
if its truth (or
falsity) is unchanged whenever any scale is replaced by another
acceptable scale (i.e. obtained by an admissible scale transformation).In other words, the truth (or falsity) of the statement is independent onthe (acceptable) scale used.
General statements:-
for a ratio scale, it is meaningful to say that one thing is so-and-so
many times as big as another (e.g. C weights five times more than A);
- If f is an interval scale, then comparisons of intervals aremeaningful, e.g. f(a) f(b) is equal/greater than f(c) f(d).-
If f
is an ordinal scale, it is meaningful to say that f(a) is greater
than f(b).
Meaningfulness of a statement
-
8/13/2019 Theory for performance Indicators
27/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
27
Examples
-
the statement x1 x2 = 3(x3 x4
)
is meaningful for an intervalscale
(e.g. try with the linear transformation (x) = ax + b, being a>0),but not for an ordinalscale (e.g. try with the monotonic transformation(x) = x2);
- the statement x1 > x2 + x3 + x4 is meaningful just for a ratioscale(e.g. try with the similarity transformation (x) = ax, being a>0).
When a transformation of scale generates a paradox, it means that thestatement is not permissible (for that scale).
How to check the meaningfulness of a statement?
-
8/13/2019 Theory for performance Indicators
28/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
28
Try by yourself to check these:
-
the statement that your telephone number is twice as large as mine is
meaningless, because a change by one number in the last digit of
your
phone number
certainly an admissible transformation
could change
the statement from a true one to a false one.
- the statement that you weight less than the Statue of Liberty is true,independently of what scale is used to measure weight, and hence is a
meaningful statement.
Other Examples
-
8/13/2019 Theory for performance Indicators
29/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
29
(x) = (9/5)x + 32
yesterday today waters triple point
Temp. in C 5 10 0
Temp. in F 41 50 32
Today the temperature is twice as yesterday.
Today the difference between the air temperature and thetemperature of the waters triple point (i.e. =0C) is twice as
yesterday.
-
8/13/2019 Theory for performance Indicators
30/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
30
Using the previous checking mechanism, we can provide an indirect
demonstration of what reported in the Tab. 1, about the permissiblestatistics:
Example
Statement: the mean value of three objects is greater/smaller/equal toa certain quantity (defined on the scale of interest).
It is meaningful for an intervalscale (e.g. try with the lineartransformation (x) = ax
+ b, being a>0);
but it is not meaningful for an ordinalscale (e.g. try with themonotonic transformation (x) = x2). Instead, the median ismeaningful for ordinal scales
Meaningfulness of a statement about the use of(permissible) statistics
WARNING:It is not appropriate to make statements using non-permissible statistics
-
8/13/2019 Theory for performance Indicators
31/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
31
1
2
3
45
6
7
8
9
10
1112
0 1 2 3 4 5 6 7 8 9 10
Example 1
1 2 3 4 5 6 7 8
For which scale(s) is meaningful to say that ? 3x x
x1 x2 x3 x4 x5x
Data set: x1 = 1, x2 = 2, x3 = 4, x4 = 5, x5 = 8
1 2 3 4 5 6 7 10 11
x1 x2 x3 x4 x5
x
Lets try with a (monotonically increasing) transformation,
permissible for ordinal scales
(TRANSFORMED SCALE):
New data set: x1= 2, x2= 3, x3= 5, x4= 6, x5= 11
(x)
-
8/13/2019 Theory for performance Indicators
32/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
32
1
2
3
45
6
7
8
9
10
1112
0 1 2 3 4 5 6 7 8 9 10
Example 2
1 2 3 4 5 6 7
For which scale(s) is meaningful to say that ? 3x x
x1 x2 x3 x4 x5x
Data set: x1 = 1, x2 = 2, x3 = 4, x4 = 5, x5 = 6
1 2 3 4 5 6 7 8 9 10
x
Lets try with a (monotonically increasing) transformation,
permissible for ordinal scales
(TRANSFORMED SCALE):
x 3.6
x ' 5.2
x1 x2 x3 x4 x5
New data set: x1= 2, x2= 3, x3= 5, x4= 6, x5= 10
(x)
-
8/13/2019 Theory for performance Indicators
33/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
33
(Order) Categories C1 C2 C3 C4 C5
Occurrences 1 3 3 1 1
Scale 1 1 2 3 4 5
Scale 2 1 4 9 16 25
(x) = x2
MEDIAN MEAN
3 2.78
9 (= 32
) 9 (2.782
)
In detail (individual measurements): sum median mean
Scale 1 1, 2, 2, 2, 3, 3, 3, 4, 5 25 3 25/9 = 2.78
Scale 2 1, 4, 4, 4, 9, 9, 9, 16, 25 81 9 81/9 = 9
confirmation:
the mean is not an admissible statistic for ordinal scales
try yourself with other statistics/scales!
A similar case
-
8/13/2019 Theory for performance Indicators
34/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
34
What about the indicators of dispersion?
-5 -4 -3 -2 -1 0 1 2 3 4 5
N 28
x 0.5
x 1R 10
s 3.1xx
(SCALE 1)
1 2 3 4 5 6 7 8 9 10 11
N 28x ' 5.5 (x)
x ' 5 (x)
R' 10 R
s' 3.1 s
(SCALE 2)
Redefine the "zero point" on this interval scale:x = (x) = 1x + 6
sR
x 'x '
s'R'
-
8/13/2019 Theory for performance Indicators
35/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
35
What about the indicators of dispersion?
-5 -4 -3 -2 -1 0 1 2 3 4 5
N 28
x 0.5
x 1R 10
s 3.1
(SCALE 1)
-4 -2 0 2 4 6 8 10 12 14 16
N 28x '' 5 (x)
x '' 4 (x)
R'' 20 2 R
s'' 6.2 2 s
(SCALE 3)
Full
linear transformation (redefinition of the "zero point
and rescaling):
x = (x) = 2x + 6
xx
sR
x ''x ''
s''R''
-
8/13/2019 Theory for performance Indicators
36/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
36
1
2
3
45
6
7
8
9
10
1112
0 1 2 3 4 5 6 7 8 9 10
Example 3
1 2 3 4 5 6 7
For which scale type(s) is meaningful to say that ?1 1 2 3 2 4 5R (x ,x ,x ) R (x ,x )
x1 x2 x3 x4 x5x
Data set: x1 = 1, x2 = 2, x3 = 4, x4 = 5, x5 = 6
1 2 3 4 5 6 7 8 9 10
x
Lets try with a (monotonically increasing) transformation,
permissible for ordinal scales
(TRANSFORMED SCALE):
x1 x2 x3 x4 x5
New data set: x1= 2, x2= 3, x3= 5, x4= 6, x5= 10
(x)
1R 2R
1R ' 2R '
-
8/13/2019 Theory for performance Indicators
37/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
37
Example 4Provide an analytical demonstration of the meaningfulness of the
previous
statement for interval-
and ratio-scales.
Try the same with the statement 1 1 2 3 2 4 5(x ,x ,x ) (x ,x )
-
8/13/2019 Theory for performance Indicators
38/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
38
Show that the geometric mean, i.e.
is an admissible central tendencyindicator for ratio-scales
only.
Show that the harmonic mean, i.e.
is an admissible central tendencyindicator for ratio-scales
only.
Some additional exercises for you!1/ n
n
nGM i 1 2 n
i 1
x x x x ... x
HM n
1 2 ni 1 i
n nx
1 1 11...
x x xx
-
8/13/2019 Theory for performance Indicators
39/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
39
The concept of Measurement according to theRepresentational theory of Measurement (by Roberts)
We now study the concept of Measurement
from a more
theoretical and philosophical perspective.The modern theory of measurement is representational: numbersassigned to objects/events must represent the perceived relations
between the properties of those objects/events.The representational definition is based on four parts:
1) An empirical relational system;
2) A symbolic/numerical relational system,
3) A representation condition;
4) A non-uniqueness condition.
-
8/13/2019 Theory for performance Indicators
40/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
40
1) Empirical Relational System
Consider a specific characteristic/property/feature (for example thelength of an object) of a well-defined context/system.
A is the set of all possible manifestationsof the characteristic:
R is a family of empirical relationsamong the manifestations of A:
then is called empirical relational system
1
A ,..., ,...i
a a
1R ,..., mR R
A,RA
-
8/13/2019 Theory for performance Indicators
41/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
41
2) Numerical/Symbolic Relational System
Analogously, N is a set of symbols (generally numbers):
and P is a family of relationsamong the elements of N:
Then is called symbolic/numerical relationalsystem
1N ,..., ,... in n
1P ,..., mP P
,PZ
-
8/13/2019 Theory for performance Indicators
42/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
42
3) Representation ConditionIn general, according to the so called Symbolic RepresentationalTheory, a measurement is an objective empirical
function which
maps homomorphicallythe empirical relational system into thesymbolic/numerical relational system
P
NA
RF
A Z
a3a2
a1 n1
n2
R2
R1
P2
P1
-
8/13/2019 Theory for performance Indicators
43/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
43
2 mappings are defined:
1) M: A N Homomorphism (homo=alike, but not identical)its not
a one-to-one mapping (separate manifestations/entities may beindistinguishable from the viewpoint of the feature of interest,
e.g. the
height of an individual)
2)
F: R
P Isomorphism
(i.e., samestructure)
its a one-to-one
mapping.
Empirical relations should be preserved by symbolic/numerical relations
In other words, whatever inference can be made in the numericalrelational system MUST apply to the empirical one (e.g., you are tallerthan the queen of England
or your weight is twice as much as mine).
3) Representation Condition
-
8/13/2019 Theory for performance Indicators
44/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
44
The representation condition may be valid for more than one mappingfunction M (conversely, the second mapping (F) is unique).
There are admissible transformations from one scale to anotherwithout invalidating the representation condition.The uniqueness condition defines the class of transformations
for
which the representation condition is valid.
E.g., for ordinal scales, all monotone increasing functions areadmissible transformations.
E.g., regarding dimensional measurements (defined on ratio scales), a
plethora of scales based on different units are commonly used(meters, foots, inches, miles, (nautical) leagues, etc).
Pounds to grams: (x)= 453.6 x
4) Non-Uniqueness Condition
-
8/13/2019 Theory for performance Indicators
45/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
45
Definition of indicator according to theRepresentational Theory
An indicator ( I ), similarly to a measurement, can beconsidered as a map from an empirical system (the realworld) into a symbolic system (usually, a numerical system).However, the mapping between the empirical and symbolrelations (ISOMORPHISM), unlike measurement, is notrequired.
the result is that some relations among numbers MAY NOTreflect the empirical relations among manifestations.
In addition, the mapping between A and N should not
necessarily be objective.In this sense, the concept of Indicator is more relaxed.
-
8/13/2019 Theory for performance Indicators
46/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
46
The Representational Theory of Indicators
P
A
RF
IA Z
a3
a2
a1 n1
n2
R2
R1
P2
P1
: A N i i iI a I a n
-
8/13/2019 Theory for performance Indicators
47/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
47
No isomorphism among empirical andsymbolic/numerical relations
Ia1a2
a4
a5a3 n
1
n2
A N
In general, for indicators, the mapping of the empirical systeminto a symbolic may introduce new relations (not present in the
empirical system) or modify the existing ones.
-
8/13/2019 Theory for performance Indicators
48/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
48
INDICATORS MEASUREMENTS
-
8/13/2019 Theory for performance Indicators
49/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
49
A context
is a part of reality that we are focusing on (e.g. a
manufacturing process or a logistic process).
A representation-target is a specific aspect of a certain context that werepresent by one or more indicators (e.g., in the case of a logisticprocess, the classification of suppliers, the delivery time etc).
A model
is a conceptualization of the representation-target, obtained
by defining one or more indicators.
Some definitions
EmpiricalSystem
Model
Symbolic/NumericalSystem
INDICATOR(S)DEFINITIONINTERPRETATION AND DEFINITION OF
THE REPRESENTATION TARGET
-
8/13/2019 Theory for performance Indicators
50/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
50
Example 4.3 The wheelbase of a motor vehicle (the geometricaldistance between two car-axles) is a dimensional measurement, and
therefore also an indicator. The relationships among symbolicmanifestations (numerical distance values) are isomorphically
linked to
the relations among physical manifestations (physical distance).
Example 4.6 Let us consider the representation-target classification
of the students of a class, operationalized (i.e. modeled) by theindicator name of the student. This indicator associates each student(empirical manifestation) to the corresponding name (symbolicmanifestation). In nature there is no order relation among the empirical
manifestations (the students), which corresponds to the alphabeticalorder relation among the symbolic manifestations (names). So, thename of the student is only an indicator, not a measurement.
In other words: the order relation among symbolic manifestations doesnot correspond to any existing relation among real manifestations.
Indicators and measurements
PERFORMANCE INDICATORS b D M i
-
8/13/2019 Theory for performance Indicators
51/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
51
Given a representation-target, the related indicator (or set ofindicators) is not univocally defined.
The same representation-target can be represented by moreindependent indicators (or indicator sets).
Example 1: Let us consider 4 production lines .
The representation-target is identifying the line with the lower defectiveness. Atleast, two different indicators can be adopted:
1) The number of defective products in the daily production;
2) The number of defects detected during a full control of the daily production.
Of course, there is no mathematical transformation which biunivocally
links the
two indicators.
Example 2: different indicators about to evaluate job stress (e.g., NIOSH
equation, Strain Index, Ergonomitermometer, and many others).
Non-uniqueness for (single) indicators
, , ,
PERFORMANCE INDICATORS b D M i
-
8/13/2019 Theory for performance Indicators
52/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
52
Non-uniqueness for (single) indicators
z''1
z''2
z'''1
z'''2
z'''3
I
I
I
Real manifestations ofthe empirical system
Manifestations of different
symbolic systems
a1
a2
a4
a5
a3
z'1
z'2 z'3
z'4
A
Z
Z
Z
Fig. 4.2.Schematic representation of the condition of non-uniqueness. The same
representation target is operationalized by three different indicators (I, I andI).Some empirical manifestations, indistinguishable according to one indicator,
can be distinguished by another one (for example the manifestations a3and a5 are
undistinguished byI, but distinguished byIandI)
PERFORMANCE INDICATORS b D M i
-
8/13/2019 Theory for performance Indicators
53/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
53
Measurements the requirement of homomorphism for mappingempirical manifestations and isomorphism for mapping relations definesa class of equivalent scales. Each equivalent scale can be mappedinto another. All the possible transformations form the so called class
of admissible transformations.
Indicators
there can be different indicators (or sets of indicators)
which represent the same representation-target. However, there maynot necessarily be a transformation
from one indicator into another.
Analogous representation-targets might not be comparable, ifrepresented by different indicators.
Non-uniqueness: comparison between measurement andindicators
PERFORMANCE INDICATORS by D Maisano
-
8/13/2019 Theory for performance Indicators
54/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
54
Different indicators sets may refer to the same representation-target.What is the best way of selecting them?; When is the representation
exhaustive?.Even if there are many ways for representing the same process, thebest
one cannot be a-priori identified.
We will see some properties and rules to support the selection andaggregation of indicators and the verification of the representationmodel.
Non-uniqueness for Sets of indicators
PERFORMANCE INDICATORS by D Maisano
-
8/13/2019 Theory for performance Indicators
55/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
55
Example 4.2 In a manufacturing company producing hydraulic valves,the purpose is to improve the quality of the produced goods. The
following indicator set is implemented to operationalize thisrepresentation target:- I1 number of units produced;
- I2
(monthly) number of units categorized as defective, and rejected.
A second possible indicators set is given by:
- I1
number of units produced by the first of 4 production lines;
- I2
average percentage of detective units: result of a spot check onthe 5% of total production.
Non-uniqueness for Sets of indicators
PERFORMANCE INDICATORS by D Maisano
-
8/13/2019 Theory for performance Indicators
56/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
56
Objective indicators. They link empirical manifestations to symbolicmanifestations objectively. The mapping does not depend on the
subject who performs it.
Example 4.7 Let us consider the indicator: quantity of goodsproduced in a plant. The empirical manifestation (production) canobjectively be connected to a corresponding symbolic manifestation(number of goods produced). If there is no counting mistake, differentpeople (or automatic devices even) will determine the same value.
Objective indicators are not necessarily measurements (think ofExample_4.6 about the name of a student).
objectivity is a necessary but not sufficient condition to be a
measurement.
Classification: Objective and Subjective indicators
PERFORMANCE INDICATORS by D Maisano
-
8/13/2019 Theory for performance Indicators
57/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
57
Subjective indicators. Empirical manifestations are subjectively mappedinto symbolic manifestations, depending on subjective perceptions or
personal opinions (different people can map the same empiricalmanifestation into different symbolic manifestations).
Example: the individual vote at the elections.
Example 4.8 The representation-target evaluation of the design of acarcan be operationalized
by the indicator quality of the design,
codified with a 5 level scale (1-very bad; 2-poor; 3-fair; 4-good; 5-
excellent).The indicator is subjective because the same empirical manifestation (aparticular car design) can be associated to different symbolicmanifestations (the 5 scale levels), depending on the subject.
Classification: Objective and Subjective indicators
PERFORMANCE INDICATORS by D Maisano
-
8/13/2019 Theory for performance Indicators
58/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
58
Measurements as specific Objective indicators
reference for
comparison
object to be
measured
(metrological)traceability
chain
PERFORMANCE INDICATORS by D Maisano
-
8/13/2019 Theory for performance Indicators
59/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
59
Traceability chain (it guarantees Objectivity)
PERFORMANCE INDICATORS by D Maisano
-
8/13/2019 Theory for performance Indicators
60/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
60
Subjective indicators: the reference is in the mind of theevaluator
Lev.
reference
referenceperformance levels evaluation
scale
subjective
judgment
comparisonterm
object to be
evaluated
Lev. Lev. Lev.Lev.
PERFORMANCE INDICATORS by D Maisano
-
8/13/2019 Theory for performance Indicators
61/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
61
Since subjective indicators provide essential information about theindividuals behaviour and perceptions, they are often used and studied
by many disciplines in the area of Social, Behavioural and CognitiveSciences.The numeric encoding is a common way to make the informationpractical for the user. However
when the relations among symbolic
manifestations do not correspond to relations among empirical ones this sort of conversion may distort the analysis results
Sometimes, the symbolic relations are (questionably) promoted,with reference to the empirical ones.
The importance of Subjective indicators
PERFORMANCE INDICATORS by D.Maisano
-
8/13/2019 Theory for performance Indicators
62/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
62
Importance levels
related to the Customer Needs within the HoQ.
they are supposed to be defined upon an ordinal
scale; however
their distance matters!Coefficients
of the relationship matrix (, , ) within HoQ.
Judgments
within AHP or Evaluations
within FMECA.
just to mention some indicators familiar to you.
Examples of promotions
PERFORMANCE INDICATORS by D.Maisano
-
8/13/2019 Theory for performance Indicators
63/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
63
Basic indicators. They are obtained from a direct observation of ageneric system.
Derived indicators. They are obtained combining the information of oneor more source
indicators (basic or derived). Synonyms: composite,
aggregatedindicators.
Example 4.9 Let consider the derived indicator:percentage ofdefectivesin a production line, given by:
Classification: Basic and Derived indicators
(number of defective units)1
3 2 (total number of produced units)
II
I
PERFORMANCE INDICATORS by D.Maisano
-
8/13/2019 Theory for performance Indicators
64/137
y
Course of QUALITY ENGINEERING
64
Example 4.10An organization for the environmental protection askstwo local Agencies -
A and B -
to estimate the pollution level of the
exhaust emissions of a motor vehicle, on the basis of four pollutantsconcentrations:
: the concentration of NOX
in the exhaust emissions [g/m3];
: the concentration of un-burnt hydrocarbons [g/m3];
: the concentration of CO [g/m3];
: the concentration of (PM10
) [g/m3].
Other examples of Derived indicators
XNOI
COI
HC
I
10PMI
PERFORMANCE INDICATORS by D.Maisano
-
8/13/2019 Theory for performance Indicators
65/137
y
Course of QUALITY ENGINEERING
65
Other examples of Derived indicators
BASIC Indicators DERIVED Indicators (I grade) DERIVED Indicator (II grade)
INOx
IHCICOIPM10
INOx (5 level scale)IHC (5 level scale)ICO (5 level scale)IPM10 (5 level scale)
Agency A:
IATOT = max {INOx,IHC,ICO ,IPM10 }
Agency B:
Real concentration of four airpollutants (expressed as g/m3) Mapping of the concentrationinto a 5 level (qualitative) scale Aggregations of the four derivedindicators into another one
10' ' ' '4
XNO HC CO PMB
TOT
I I I II
non uniqueness of theaggregation criteria
PERFORMANCE INDICATORS by D.Maisano
-
8/13/2019 Theory for performance Indicators
66/137
y
Course of QUALITY ENGINEERING
66
Derived indicators according to the Representationalapproach
Basic Indicators(I1, I2, I3)
Derived Indicator(I4)
Indicatorsaggregation
a1
a2a3
x1
x2
b1
b2
y1
y2
c1c2
c3
c4
z1
z2
z3
I1
I2
I3
A
B
C
X
Y
Z
w1
w2z3
w4
w5
I4d1:(x1, y1, z1)
d2:(x1, y1, z2)
d3: (x1, y1, z3)d4:(x2, y1, z1)
d5:(x2, y1, z2)
d6:(x2, y1, z3)
d7:(x1, y2, z1)
d8:(x1, y2, z2)
d9:(x1, y2, z3)
d10:(x2, y2, z1)
d11:(x2, y2, z2)
d12:(x2, y2, z3)
W
Sets of the empirical
manifestations of thesource indicators
Sets of the corresponding
symbolic manifestations ofthe source indicators
New set of empirical manifestations:
combination of the source indicatorssymbolic manifestations: D =(X x Y x Z)
Set of the symbolic
manifestations of thederived indicator
PERFORMANCE INDICATORS by D.Maisano
-
8/13/2019 Theory for performance Indicators
67/137
Course of QUALITY ENGINEERING
67
The problem of the aggregationThe aggregation of several indicators into a derived indicator is notalways easy to achieve, especially when the information to synthesize
is assorted.The aggregation of heterogeneous indicators is often complex and
based on questionable simplifications.Example: Analyze the derived indicator RPN (= SOD, from FMECA).
PERFORMANCE INDICATORS by D.Maisano
-
8/13/2019 Theory for performance Indicators
68/137
Course of QUALITY ENGINEERING
68
A Utopian schemeDerived indicators may be (in turn) aggregated into a derived indicatorof higher grade.
By extending this concept to the limit, we can imagine to define asuper-indicator, synthesizing all the aspects of the systeminvestigated.
I1
I2
I3
I4I5
BASIC Indicators DERIVED Indicators(I grade aggregation)
DERIVED Indicator(II grade aggregation)
Starting indicators
Super-indicator ofglobal performance
PERFORMANCE INDICATORS by D.Maisano
-
8/13/2019 Theory for performance Indicators
69/137
Course of QUALITY ENGINEERING
69
Problem: What is the best production line?
Example of aggregation
Table 3.1. Experimental data of four equivalent production lines for exhaust-
systems in a manufacturing plant
Production lines
Indicators
daily production [no. per day] 360 362 359 358
daily defectiveness [no. per day] 35 32 36 40
unavailability equipment ratio [%] 4.00% 5.50% 4.50% 5.00%
For each basic indicator we may establish the following rankings:
daily production:
daily defectiveness:
unavailability equipment ratio:
1 2 3 4
PERFORMANCE INDICATORS by D.Maisano
-
8/13/2019 Theory for performance Indicators
70/137
Course of QUALITY ENGINEERING
70
For each line, we can associate a Bordas
indicator IB
(x):
where Ii
(x)is the ranking obtained by a line xwith regard to i-th
basic
indicator and m is the number of indicators used (in this case, m=3).
The winner (the best line x*) is given by (Borda
1781):
where Sis the set of compared lines. In this example, .
Bordas indicator
1
m
B iiI x I x
* minB Bx S
I x I x
, , ,S
PERFORMANCE INDICATORS by D.Maisano
-
8/13/2019 Theory for performance Indicators
71/137
Course of QUALITY ENGINEERING
71
Some simplifying assumptions/drawbacks:
-
the (3) different criteria have the same weight;
- equal distance between two consecutive rank positions;-
Bordas
method is sensitive to irrelevant alternatives.
Bordas indicatorApplying Bordas method to data in Table 3.1, we obtain the following
results:
2 2 1 5BI 1 1 4 6BI
3 3 2 8BI
4 4 3 11BI According to Eq. 3.16, the final ranking is: .
The winner (i.e. the line with best overall performance) is line .
PERFORMANCE INDICATORS by D.Maisano
C f QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
72/137
Course of QUALITY ENGINEERING
72
Sensitivity to Irrelevant alternativesTable 3.3. Experimental data of three equivalent production lines for exhaust-systems in a manufacturing plant (1stcondition)
Production Lines
Indicators
daily production [no. per day] 367 350 354
daily defectiveness [no. per day] 35 30 37
1 2 3BI
3 1 4BI
2 3 5BI
Table 3.4. Experimental data of three equivalent production lines for exhaust-systems in a manufacturing plant (2
ndcondition)
Production Lines
Indicators daily production [no. per day] 367 350 345
daily defectiveness [no. per day] 35 30 33
1 3 4BI
2 1 3BI
3 2 5BI
Despite the marginal role played by , its change may influence the
winner
PERFORMANCE INDICATORS by D.Maisano
C f QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
73/137
Course of QUALITY ENGINEERING
73
In Football
group tournaments, a popular ranking system is:
3-1-0 points systems for win-draw-loss.
In Formula 1
races: the winner receives 25 points, the second place
finisher 18 points, with 15, 12, 10, 8, 6, 4, 2 and 1 points for
positions
3 through 10.
Scoring system in Moto GP races:1st=25 points, 2nd=20 points, 3rd=16 points, 4th=13 points, 5th=11 points,6th=10 points, 7th=9 points, 8th=8 points, 9th=7 points, 10th=6 points,11th=5 points, 12th=4 point, 13th=3 points, 14th=2 points, 15th=1 point.
Scoring system is even more complex for Decathlon Competition.
all these rankings are questionable! A proof is that they (periodically) change over
time.
The problem of aggregation in Sports
PERFORMANCE INDICATORS by D.Maisano
C f QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
74/137
Course of QUALITY ENGINEERING
74
How to select the best indicator or (set of indicators) for a specific goal?
God only knows
However analyzing the indicator properties may
help.
Properties can be classified into four groups:
1 - general properties,2 -
properties of sets
of indicators,
3 -
properties of derived
indicators,
4 - accessory properties.These properties can represent a useful tool to select and evaluate
performance indicators in different contexts.
Properties of indicators
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
75/137
Course of QUALITY ENGINEERING
75
Properties of indicatorsCategory Properties Short description
Consistency with
the representation-
target
The indicator should properly represent the
representation-target.
Level of detail The indicator should not provide more than therequired information.
Non counter-
productivity
Indicators should not create incentives for counter-
productive acts.
Economic impact Each indicator should be defined considering the
expenses to collect the information needed.
General
properties
Simplicity of use The indicator should be easy to understand anduse.
Exhaustiveness Indicators should properly represent all the system
dimensions, without omissions.
Non redundancy Indicators set should not include redundant
indicators.
Monotony The increase/decrease of one of the aggregatedindicators should be associated to a corresponding
increase/decrease of the derived indicator.
Properties of
sets of
indicators
S = {Ii, Ij, Ik}
Properties ofderived
indicators
(Ii, Ij, Ik)ITOT Compensation Changes of different aggregated indicators may
compensate each other, without making the
derived indicator change.
Long term goals Indicators should encourage the achievement ofprocess long-term goals.Accessory
properties Impact on the
stakeholders
For each indicator the impact on process
stakeholders should be carefully analysed.
1)
2)
3)
4)
sub-
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
76/137
Course of QUALITY ENGINEERING
76
1.1 Consistency with the representation-target
Every indicator should properly operationalize
a representation-target.
The mapping should be thoroughly verified before using the indicator.
Example 4.13 Referring to the representation-target sales of a
manufacturing company, the indicator IS
total number of goods soldin the whole year
is used to represent the process. Later, companymanagers realize that quarterly information on sales would be moreuseful for estimating the seasonal trend. Consequently, a new indicator
IS
representing the total number of quarterly sold goodsreplaces the
first one.
According to the representation-target, the second indicator is moreaccurate than the first one. It comprehends some important empirical
manifestations (quarterly information on sales), ignored by IS
.
1. General properties (referred to single indicators)
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
77/137
Course of QUALITY ENGINEERING
77
1.2 Level of detail (resolution)
If an indicator maps two empirical manifestations
not distinguished
according to a representation-target into different symbolicmanifestations, then the level of detail is excessive.
Excessive level could complicate the analysis and could be economicallywasteful.
Example 4.14A manufacturing company produces metal screws. Therepresentation-target is the production rate of the company. The indicator Irepresents the daily weight of produced screws. If the indicators accuracy is1 g/day
when 10 kg/day would be enough
then the level of detail is
excessive.
Example 4.15
1. General properties (referred to single indicators)
your impression on the external design of car A is:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15VERY BAD EXCELLENT
* * * * *
your impression on the external design of car B is:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15VERY BAD EXCELLENT
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
78/137
Course of QUALITY ENGINEERING
78
1.3 Non counter-productivity
The idea of counter-productivity can be shown as follows. Some
indicators (Ik
, Ih
, Ii
, Il
, Im
, In
, Io
, Ip
) are aggregated in a derived
indicator (ITOT
), representing the global performance. If the increase of
a specific source indicator (Ik
) is associated with the decrease of one or
more indicators (for example Ih
, Ii
, Il
), determining a decrease of the
global performance (ITOT ) too, then Ik is counter-productive.
1. General properties (referred to single indicators)
Ik Ih Ii Il Im In Io Ip
increasing indicator(counter-productive)
correlated indicatorswhich inevitably decrease
indicators whichdo not change
a'
a
global
performanceITOT(a) < ITOT(a)
= = = =
obviously, theseindicators should bedefined on scales with- at least- order relation
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
79/137
Course of QUALITY ENGINEERING
79
1.3 Non counter-productivity
Example 4.16 The main purpose of a construction company is to reduce
the construction work time, in order to take a competitive advantage. Thispurpose may generate some counterproductive actions:
-
to save time, employees do not obey safety rules (i.e. they do not use the
safety helmets and harness);
- working vehicles, rushing around the building site to save time, becomedangerous for the public safety;-
customer satisfaction decreases, because the result of the work
is poor, due to
the excessive speed up.
In this case, focusing too much on a single dimension of the process can becounter-productive in general terms.
1. General properties (referred to single indicators)
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
80/137
Course of QUALITY ENGINEERING
80
1.3 Non counter-productivity
Example 4.17 To estimate the costumer satisfaction, a call-center uses
several indicators. Two of them are the following:I1
average number of rings before answering the phone;
I2
percentage of unanswered calls.
These two indicators can be counter-productive because employees can gamethe process answering the phone immediately and then putting the
call on hold
before starting the conversation.
Although that behaviour
increases the value of selected indicators, it is
absolutely counter-productive according to other indicators of customer
satisfaction. For example, the number of exhaustive answers, the
courtesy, the
number of queued calls etc.
In conclusion, the increase of I1
and I2
indicators could badly impact the proc-
ess, making the global customer satisfaction decrease.
1. General properties (referred to single indicators)
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
81/137
Course of QUALITY ENGINEERING
81
1.3 Non counter-productivity
Example.A car supplier (e.g. VW) offers an incentive contract to a dealerto encourage his effort in selling cars. The incentive is a bonusin case thenumber of cars sold in one month is larger than a threshold (e.g. 20 cars).
Possible counter-productive action: the dealer may tend to concentrate his effortin those months in which he realizes that there is a real chance
to get the bonus;
in the less promising
months
on the contrary
he may spare himself or even
postpone car sales.
The results are large fluctuations in the suppliers demand (this maybedangerous for bullwhip effect).
1. General properties (referred to single indicators)
1 2 3 4 5 6 7 8 [month]
[cars sold]
20 threshold
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
82/137
Course of QUALITY ENGINEERING
82
1.4 Economic Impact
The economic impact of an indicator strictly depends on the nature of the
system investigated. The impact can be studied in relative terms, by com-paring two different indicators operationalizing
the same representation-
target.
1. General properties (referred to single indicators)
Ia
Ib
Ic
mapping
Indicators operationalizing thesame representation-target
Economic Impact []
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
83/137
Course of QUALITY ENGINEERING
83
1.4 Economic Impact
Example 4.18A small company produces punched metal components. To
check the quality of the manufactured holes, two possible indicators can beused:
I1
)
diameter measurements, taken by using an accurate calliper. To
check each hole, the time needed is about 9 seconds.
I2
) the result of a (go - no go) manual testing, using a hard-gauge. Timeneeded is about 3 seconds.
Supposing that the cost for measurements is directly proportional to timespent, then indicator I2
can be considered three times more economical
than indicator I1
.
1. General properties (referred to single indicators)
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
84/137
Course of QUALITY ENGINEERING
84
1.5 Simplicity of use
Indicators that are difficult to understand and interpret, because
reference to real data has been lost, are often rejected by potentialusers
This property, as the previous one, can be studied in relative terms, bycomparing two (or more) different indicators operationalizing
the same
representation-target.
The comparison concerns the aspects related to simplicity of use
(for ex-
ample, indicators should be easy to understand, easy to use, they shouldhave a clear meaning, they should be largely accepted, etc...).
1. General properties (referred to single indicators)
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
85/137
Q
85
1.5 Simplicity of use
Example 4.19 Likewise Example 4.18, we set-up a mapping to evaluate the
simplicity of use of two indicators (I1
and I2
), considering the following criteria:
(a) technical difficulty in performing measurements;
(b) time required.
1. General properties (referred to single indicators)
(a) technical skill (b) time required simplicity of use(sum of the two level no.)
I1 2 3 (2 + 3) = 5
I2 1 1 (1 + 1) = 2
(3 levels) (3 levels)
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
86/137
86
A set or family of indicators is composed by the indicators selected torepresent a generic process. These indicators can be grouped into sub-
sets, depending on their characteristics.
2. Properties of sets of indicators
2AI
Dimension A2: empiricalmanifestations
Symbolic manifestationsof dimension A2
a1a2
a4
a5
a3
Dimensions of theprocess represented
A2
Empirical manifestations that are undistinguishable, according
to the representation-target operationalized by the indicator2A
I
DimensionA2exploded
v1
v3
v2
Fig. 4.12. Schematic representation of the concept of indicators set. For each
process dimension (A1, A2, A3, ), it is possible to define one or more indicators.All the indicators form an indicators set or family. Indicator
2AI represents the
dimension A2
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
87/137
87
Example 4.20 Three indicators represent a companys sales:
I1
) number of products daily sold;
I2
) daily turnover;
I3
) daily takings (not including the credit given).
Two possible process states are:i-th
day:
I1
(i) = 203 pcs;
I2
(i) = 4820 ;
I3
(i) = 3600
j-th
day:
I1
(j) = 178 pcs;
I2
(j) = 5680 ;
I3
(j) = 3546
Each state is a snapshot
of the process condition in a particular day.
States of a process/system
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
88/137
88
2.1 Exhaustiveness
Indicators of the set must treat all parts or aspects of the system of
interest, without omission.
2. Properties of sets of indicators
I
v1
v3
v2
b1b2
b4
b5
b3
Empirical manifestations of
the process dimension
Symbolicmanifestations
Empirical manifestations that are undistinguishable, according
to the representation-target operationalized by the indicator
(a) Incorrect definition of one indicator (b) Missing indicator for a certain dimension
I (missing Indicator) v1v3
v2
b1
b2
b4
b5
b3
v4
Empirical manifestations of
the process dimension
Symbolicmanifestations
Empirical manifestations that are undistinguishable, according
to the representation-target operationalized by the indicator
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
89/137
89
2.1 Exhaustiveness
Example 4.21 A manufacturing company producing metal components, uses
the following indicators:I1
) total number of units yearly produced;
I2
) manufacturing time;
I3
) lead times
(i.e. supply time, tool change time, etc...).
This set of indicators has been defined with the aim of differentiating the possiblesystem conditions. If two possible states, undistinguished by the previous indicators,are distinguished by a further indicator
which has previously been ignored (for
example I4
, the number of defective units produced)
then the set is not
exhaustive:
2. Properties of sets of indicators
I1 I2 I3 I4
State 1 300000 pcs 160000 h 700 h 2.1%
State 2 300000 pcs 160000 h 700 h 3.5%
inexhaustive set of indicators
exhaustive set o f indicators
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
90/137
90
2.1 Exhaustiveness
In general, exhaustiveness is not fulfilled when
one or more indicators are missing;
one or more indicators are not enough specific (e.g. too low resolution).
2. Properties of sets of indicators
Exhaustiveness is certainly the most important condition to guaranteeconsistency between indicators and the process/system represented.
A general rule to guarantee exhaustiveness is not available(non-uniqueness of process modeling by indicators).
Also, since every process is a dynamic system evolving over time,representation targets may change (for example, today the importance ofservices is much higher than 20 years ago).
Exhaustiveness has to be periodically checked.
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
91/137
91
2.2 Non redundancy
If a set (or family) of indicators (F) is exhaustive, and if it continues to
be exhaustive even when removing one indicator (Ik
), then the removedindicator is redundant.
Usually, indicators that can be deduced from other ones are redundant.The presence of redundant indicators does not provide additional
information on the process.
2. Properties of sets of indicators
if I1 I2 I3 I4 I5is an exhaustive set of
indicators
and ifI1 I2 I4 I5 is a set which continuesto be exhaustive
then I3 is a redundant indicator
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
92/137
92
2.2 Non redundancy
Example 4.22 in a manufacturing company producing plastic component, theprocess is represented by four indicators:
I1
)
total number of units (yearly) produced;
I2
)
number of defective units (yearly) produced;
I3
)
manufacturing time;
I4
)
efficiency of the production, calculated as:
(term I5
refers to lead times
(supply time, tool change time, repairing time, etc..);
I5
) lead times.
Assuming that the set of indicators fulfils the property of exhaustiveness, theindicator I3
is removed from the set. If the residual set (I1
, I2
, I4
, I5
) continues
to be exhaustive, then the indicator I3
is categorized as redundant.
2. Properties of sets of indicators
3 54
3
I III
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
93/137
93
In general these properties (Exhaustiveness and Non redundancy) areamong the most difficult to analyze.
An indirect proof is that the scientific literature includes several modelsto support the development of a system of performance indicators or
measurements.
Two of these are:
- the Balanced Scorecard Method;
- the EFQM
(European Foundation for Quality Management) model.
2. Properties of sets of indicators
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
94/137
94
The performance of a complex system is related to severalinterconnected dimensions;
Indicators that make one dimension look good while deflating anotherare avoided (minimize negative competition).
The number of measures has to be limited to a vital few (not tocomplicate or to make the system representation too expensive).
The concept of balanceconsists in placing the right emphasis on all thesystems important dimensions
(e.g. often the economic/financial dimension
is favored).
Balanced Scorecard
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
95/137
95
Balanced Scorecard
CustomerHow well do we satisfy
our internal and externalcustomers needs?
FinancialHow do we look toour stakeholders?
Internal Business ProcessHow well do we perform at
key internal businessprocesses?
Learning and Growth
Are we able to sustain innovation,change, and continuous
improvement?
Secondary Influence on PerformancePrimary Driver of Performance
Fig. 5.6. The four perspectives of the Balanced Scorecard model (Kaplan and
Norton 1992). With permission
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
96/137
96
For each dimension/perspective, a (sub-)set of performance indicatorshave to be identified.
A way to help this task is to answer the following questions:
-
Financial
What are our strategic financial objectives?
- Customer What do we have to do for our customers in order toensure our financial success?-
Internal Business Process
Which of our business processes most im-
pact customer satisfaction?
-
Learning and Growth
What improvements can be made to ensure
sound business processes and satisfied customers?
Balanced Scorecard
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
97/137
97
Balanced Scorecard: visualization by the performance dashboardTemperature
HC
Fuel
FE
MaturityHow long has our organizationbeen on the road? How old isour measurement system? Is it
time for a check-up?
0 2 5 0 0 0
Odometer
Speed
110
RPM x 1000
70
Internal Business ProcessAre our internal business processesoperating efficiently and effectively?
In what gear are we operating?
FinancialDo we have the financial
resources and stability to reachour destination? Do we have the
backing of our stakeholders?
CustomerAre we addressing and meetingthe need of our customers? Arethey cold (not participating)? Are
they hot (complaining)?
Learning and GrowthAre we growing and improving at
a sustainable pace? Are wemoving too slow? Too fast?
Fig. 5.7. The Balanced Scorecard as a performance dashboard (Performance-
Based Management Special Interest Group 2001). With permission
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
98/137
98
The EFQM model can be used to assess an organizations progress to-wards excellence, independently of the organizations type, size,
structure, and maturity.
It is based on the premise that Excellence depends on the capacity ofconciliating the stakeholders
different exigencies and interests.
The model is based on nine criteria (dimensions). Five of these areclassified as Enablers
and four as Results. The Enabler
criteria
cover what an organization does; the Result
criteria cover what an
organization achieves. Feedback from Results
help to improve
Enablers.
EFQM method
-
8/13/2019 Theory for performance Indicators
99/137
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
100/137
100
3.1 Monotony
Let us consider a set of sub-indicators aggregated by a derived
indicator.
If the increase/decrease of (the performance of) one sub-indicator isnot associated to the increase/decrease of (the performance of) thederived indicator, then the derived indicator does not
fulfil
the condition
of monotony.
This definition implicitly entails that the symbolic manifestations of thesub-indicators are represented using a scale with (at least) order
relation.
3. Properties of derived indicators
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
101/137
101
3.1 Monotony
If a process is represented by different sub-indicators aggregated into aderived indicator (I
TOT
), and if the process skips from state S to state
S*, increasing/decreasing one sub-indicator Ik
, (not changing other
indicators
performance), then ITOT
should in-crease/decrease too.
Otherwise, ITOT
is not monotonous., then the derived indicator does not
fulfill the condition of monotony.
3. Properties of derived indicators
z1
z2
symbolic manifestations ofthe derived indicatorITOT
DERIVED Indicator
(ITOT)
S:(I1=x1, I2=x3, I3=x5)
S: (I1=x1, I2=x3, I3=x7)
manifestations of the sub-indicators aggregated byITOT
Fig. 4.20.Schematic representation of the condition of monotony. If process skips
from state Sto state S*, beingI1(S*)=I1(S), I2(S
*)=I2(S), and I3(S
*)>I3(S), then
the Monotony entails thatITOT(S*)>ITOT(S)
strictly
monotonic
weakly
monotonic
f(x)
x
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
102/137
102
3.1 Monotony
Example 4.24 The pollution level of the exhaust emissions of a motorvehicle is estimated using the following model:
The example shows that using a derived indicator which is notmonotonous, we may lose some information (according to , thereis no difference between state S and state S*).
3. Properties of derived indicators
10
' ' ' 'max , , ,X
A
TOT NO HC CO PM I I I I I
'
XNOI '
HCI '
COI 10'
PMI A
TOTI
state S 1 1 1 3 3
state S 2 3 2 3 3
ATOTI
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
103/137
103
3.2 Compensation
Given a derived (or aggregated) indicator,
- if the variation of a sub-indicator (always) makes the derived indicatorvalue change,-
and if there (always) exists another sub-indicator which variation may
cancel the variation in the derived indicator,
then the derived indicator fulfills the property of full compensation
and
a substitution rate
among sub-indicators can be calculated.
3. Properties of derived indicators
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
104/137
104
3.2 Compensation (in other words)
SI1
and SI2
are two sub-indicators
aggregated by a derived indicator D.
- If a variation SI1 (always) makes D change (D),-
and if there (always) exists a variation SI2
, so that D
is
compensated,
D
fulfills the property of full compensation,
and the function which relates SI1
and SI2
is defined as substitution
rate
between SI
1 and SI
2 .
Compensation is a typical property of additiveand multiplicativemodels.
3. Properties of derived indicators
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
105/137
105
3.2 Compensation
Example 4.25 The pollution level of motor vehicle exhaust emissions isestimated by :
As illustrated in the Table, the pollution level skips from state S to stateS*. The decreases of and are compensated by the increase of
value does not change.
3. Properties of derived indicators
10' ' ' '
4 X
NO HC CO PMB
TOTI I I II
XNO
I HCI COI 10PMI
BTOTI
state S 2 2 1 3 (2+2+1+3) / 4 = 2state S* 1 1 3 3 (1+1+3+3) / 4 = 2
XNO
IHC
ICO
I
BTOTI
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
106/137
106
3.2 Compensation
Substitution rate:
It is the portion x1 of one sub-indicator x1
that you should give up in
order to obtain an increase x2
in a second sub-indicator x2
, keeping the
aggregated indicator (y) constant.
Example: y = 2x1 + x2 + x3 x2
= -2x1
In general, if y is a continuous and derivable function with respect tothe sub-indicators, the substitution rate can be obtained from the
calculation of the infinitesimal increments (slope) of the indifferencecurve:
3. Properties of derived indicators
22 1
1
x2 x 2 x
x
x1
x2
x2
x1
y=constant
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
107/137
107
3.2 (Partial) Compensation
Example 4.24 The pollution level of the exhaust emissions of a motor
vehicle is estimated using the following model:
In case (b), no variation of other sub-indicators may compensate forthe resulting variation in the derived indicator.
A substitution rate cannot be defined univocally.
3. Properties of derived indicators
10
' ' ' 'max , , ,X
A
TOT NO HC CO PM I I I I I
'XNO
I 'HCI
'
COI 10'
PMI A
TOTI
state S 1 1 2 3 3
state S 4 1 2 3 4
'
XNOI '
HCI '
COI 10'
PMI A
TOTI
state S 1 1 2 3 3
state S 1 1 2 2 2
'
XNOI '
HCI '
COI 10'
PMI A
TOTI
state S 1 1 2 3 3
state S 4 1 2 3 4
(a) (b)
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
108/137
108
Two properties are introduced to help identifying representation
targets
which are consistent with the strategic objectives.
We underline that the properties are defined accessory because theyare helpful for testing process representation targets, rather thanindicators.
4.1 Long term goals:
Indicators should encourage the achievement of process
long-term
goals. Therefore, representation-targets should concern process
dimensions which are strictly linked to these goals.4.2 Impact on the Stakeholders/Customer orientation:
Many indicators focus on internal needs such as throughput, staffefficiency, cost reduction, and cycle time. While these needs are alllaudable, they usually have little direct impact on costumers needs. So,
it is important to identify process aspects with a strong impact oncustomer satisfaction.
4. Accessory properties
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
109/137
109
1. Replenishment lead time2. On-time performance3. Supply flexibility4. Delivery frequency5.
Supply quality
6. Inbound transportation cost
4.1 Long-term goals
Example: Supplier Scoring Assessment
a) Short-term indicator: focusing only on the quoted price.
b) Long-term indicator: it should take into account and aggregate
other important dimensions:
7. Pricing terms
8. Information coordinationcapability
9. Design collaboration capability
10. Exchange rates, taxes11. Supplier viability
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
110/137
110
Definition and testingof Indicators
Preliminary definition of indicators
Testing of the other indicators general properties
Indicators correctionor redefinition
Indicators validation
Do indicators fulfil theproperties? YESNO
Process identification
Representation-targetsidentification
Do representation-targetsfulfil the accessory
properties?YESNO
Correction ofrepresentation-targets
Check of the consistency with therepresentation-target (for each indicator)
Check of exhaustiveness and non redundancy
Check of the derived indicators properties
Several recursive steps arenecessary, before developing aproper model.
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
111/137
111
Measuring/evaluating the customer satisfactionis essential for Quality.
Understanding the customer behavior, attitude or perceptions is also
fundamental when designing a product/service.The fact that evaluations are very often subjective
does not imply that
they can be neglected.
they are studied by many disciplines in the area of Social, Behavioural
and
Cognitive Sciences.
The scientific literature shows plenty of techniques to design scales,
questionnaires, surveys, and analyze the relevant results (in a statisticallysound manner).
The importance of indicators for Quality
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
112/137
112
This activity is fundamental in marketing.
Understanding the attitude of customers is important to predict -
or
even influence! - their (future) behavior.From psychology:
-An attitudeis a residuum of experience, by which further activity is
conditioned and controlled. We may think of attitudes as acquiredtendencies to act in specific ways toward objects(Krueger & Reckless,1931).
-Attitude can be defined as readiness to respond to apsychological
objectwith some degree of (dis)favorableness(Thurstone, 1932).In general, the evaluative reaction of favor or disfavor can range from
extremely negative to extremely positive, through the neutral point, on a
dimension such as: good bad, pleasant unpleasant, or in favor opposed.
Measuring the Attitude
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
113/137
113
From psychology:
They are simply the things that psychologists take to be their properobjects (i.e., attributes, concepts) of investigation.
Examples of psychological objects: learning capability, motivation,intelligence, customer (un)satisfaction
with a service, your opinion
about smoking cigarettes, etc
Psychological Object
The attitude can be measured by generating items (i.e., statements orquestions) about the psychological object and analyzing theanswer/opinion
of the user.
basically, an indicator with a corresponding scale of measurement
How to measure the Attitude?
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
114/137
114
Psychological object:
Sarah
PalinOn a scale of 10 to +10, how would you rate Sarah Palin as a potentialPresident of US?
where:
10 = Strongly disapprove
+ 10 = Strongly approve
0 = Neutral; neither approve nor disapprove
On the same scale, how would you rate the Sarah Palins opinion onabortion?
General Question: How to define and organize these items properly?
Example of possible Items
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
115/137
115
Likert Scale (rating scale)
A Likert
scale is an ordinal scale format that asks respondents to
indicate the extent to which they agree or disagree with a series ofmental or behavioral belief statements about a given (psychological)object.
Example:
Scaling Techniques
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
116/137
116
Likert Scale (it continues)
-The analysis can be conducted on an item-by-item basis (profile
analysis), or a total (summated) score can be calculated.-When arriving at a total score, the categories assigned to the negativestatements by the respondents should be scored by reversing thescale.
Scaling Techniques
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
117/137
117
Semantic Differential scale (rating scale)
A semantic differential scale is unique bipolar ordinal scale format that
captures a persons attitudes and/or feelings about a given(psichological) object.Example:
Scaling Techniques
PERFORMANCE INDICATORS by D.Maisano
Course of QUALITY ENGINEERING
-
8/13/2019 Theory for performance Indicators
118/137
118
Semantic Differential scale (it continues)
-The negative adjective or phrase sometimes appears at the left side of
the scale and sometimes at the right.
This controls the tendency of some respondents