Sei sulla pagina 1di 13

282 PART II • RESEARCH DESIGN FORMULATION

Apple was the world's


and America's most
admired company in 2008.

Measurement and Scaling


measurement Measurement means assigning numbers or other symbols to characteristics of objects according
The assignment of numbers to certain prespecified rules. 2 Note that what we measure is not the object, but some characte1istic
or other symbols to of it. Thus, we do not measure consumers---only their perceptions, attitudes, preferences, or other
characteristics of objects relevant characteristics. Jn marketing research, numbers are usually assigned for one of two
according to certain reasons. First, numbers permit statistical analysis of the resulting data. Second, numbers facilitate
prespecified rules.
the communication of measurement rules and results.
The most important aspect of measurement is the specification of rules for assigning numbers
to the characteristics. The assignment process must be isomorphic: There mnst be one-to-one
correspondence between the numbers and the characteristics being measured. For example,
the same dollar figures are assigned to households with identical annual incomes. Only then can
the numbers be associated with specific characteristics of the measured object, and vice versa. In
addition, the rules for assigning numbers should be standardized and applied uniformly. They
must not change over objects or time.
scaling Scaling may be considered an extension of measurement. ScaJing involves creating a
The generation of a continuum upon which measured objects are located. To illustrate, consider a scale from 1 to
continuum upon which 100 for locating consumers according to the characteristic "attitude toward department stores."
measured objects are Each respondent is assigned a number from I to I 00 indicating the degree of (un)favorable-
located.
ness, with t = extremely unfavorable, and 100 extremely favorable. Measurement is the
actual assignment of a number from ] to 100 to each respondent. Scaling is the process of
placing the respondents on a continuum with respect to their attitude toward department
stores. In the opening example of most admired companies, the assignment of numbers to
reflect the annual revenue was an example of measurement. Tbe placement of individual com-
panies on the annual revenue continuum was scaling.

Scale Characteristics and Levels of Measurement


All the scales that we use in marketing research can be described in terms of four basic charac-
teristics. These characteristics are description, order, distance, and origin, and together they
define the level of measurement of a scale. 1l1e level of measurement denotes what properties of
an object the scale is measuring or not measuring. An understanding of the scale characteristics
is fundamental to understanding the primary type of scales.
CHAPTER 8 • MEASUREMENT AND SCALING: FUNDAMENTALS AND COMPARATIVE SCALING 283

Description
description By description, we mean the unique labels or descriptors that are used to designate each value of
The unique labels or the scale. Some examples of descriptors are as follows: 1. Female, 2. Male; l = Strongly
descriptors that are used disagree, 2 = Disagree, 3 Neither agree nor disagree, 4 Agree, and 5 = Strongly agree; and
to designate each value of the number of dollars earned annually by a household. To amplify, Female and Male are unique
the scale. All scales possess descriptors used to describe values 1 and 2 of the gender scale. It is important to remember that
description.
all scales possess this characteristic of description. Thus, all scales have unique labels or descrip-
tors that are used to define the scale values or response options.

Order
order By order, we mean the relative sizes or positions of the descriptors. There are no absolute values
The relative sizes or associated with order, only relative values. Order is denoted by descriptors such as "greater
positions of the descriptors. than," "less than," and "equal to." For example, a respondent's preference for three brands of
Order is denoted by athletic shoes is expressed by the following order, with the most preferred brand being listed first
descriptors such as aud the least preferred brand last.
greater than, less than,
and equal to.
Nike
Reebok
Adidas

For this respondent, the preference for Nike is greater than the preference for Reebok.
Likewise, the preference for Adidas is less than the preference for Reebok. Respondents who check
the same age category, say 35 to 49, are considered to be equal to each other in terms of age, and
greater than respondents in the 20 to 34 age group. All scales do not possess the order characteris-
tic. In the gender scale ( l. Female, 2. Male) considered earlier, we have 110 way of determining
whether a female is greater than or less than a male. Thus, the gender scale does not possess order.

Distance
distance The characteristic of distance means that absolute differences between the scale descriptors are
The characteristic of known and may be expressed in units. A five-person household has one person more than a four-
distance means that person household, which in turn has one person more than a three-person household. Thus, the
absolute differences following scale possesses the distance characteristic.
between the scale
descriptors are known and Number of persons living in your household _ _ _ _ _ _ _ __
may be expressed in units.
Notice, that a scale that has distance also has order. We know that a five-person household is
greater tlrnn the four-person household in terms of the number of persons living in the household.
Likewise, a three-person household is less than a four-person household. Thus, distance implies
order but the reverse may not be true.

Origin
origin The origin characteristic means that the scale has a unique or fixed beginning or true zero point.
The origin characteristic Thus, an exact measurement of income by a scale such as: What is the annual income of our house-
means that the scale has a hold before taxes? $ has a fixed origin or a true zero point. An answer of
unique or fixed beginning zero would mean that the household has no income at all. A scale that ha5 origin also has distance
or true zero point.
(and order and description). Many scales used in marketing research do not have a fixed origin or
true zero point, a<; in the disagree-agree scale considered earlier under description. Notice that such
a scale was defined as l = Strongly disagree, 2 Disagree, 3 = Neither agree nor disagree, 4 =
Agree, and 5 Strongly agree. However, I is an arbitrary origin or starting point. This scale could
just as easily have been defined as O = Strongly disagree, l Disagree, 2 Neither agree nor dis-
agree. 3 Agree, and 4 = Strongly agree, with Oas the origin. Alternatively, shifting the origin to
-2 will result in an equivalent scale: -2 Strongly disagree, l = Disagree, 0 = Neither agree
nor disagree, 1 = Agree, and 2 = Strongly agree. All these three forms of the agree-disagree scale,
with the origin at l, 0, or -2, are equivalent. Thus, this scale does not have a fixed origin or a true
zero point and consequently does not possess the characteristic of origin.
284 PART II • RESEARCH DESIGN FORMULATION

You may have observed that description, order, distance, and origin represent successively
higher-level characteristics, with origin being the highest-level characteristic. Description is the
most basic characteristic that is present in all scales. 1f a scale has order, it also has description.
If a scale has distance, it also has order and description. Finally, a scale that has origin also has
distance, order, and description. Thus, if a scale has a higher-level characteristic, it also has all
the lower-level characteristics. However, the reverse may not be tme, i.e., if a scale has a lower-
level characteristic, it may or may not have a higher-level characteristic. With an understanding
of scale characteristics, we are ready to discuss the primary type of scales.

Primary Scales of Measurement


There are four primary scales of measurement: nominal, ordinal, interval, and ratio. 3 These
scales are illustrated in Figure 8. I, and their properties are summarized in Table 8.1 a11d
discussed in the following sections.

Nominal Scale
nominal scale A nominal scale is a figurative labeling scheme in which the numbers serve only as labels or tags
A scale whose numbers for identifying and classifying objects. The only characteristic possessed by these scales is
serve only as labels or tags description. For example, the numbers assigned to the respondents in a study constitute a nominal
for identifying and scale. When a nominal scale is used for the purpose of identification, there is a strict one-to-one
classifying objects. When correspondence between the numbers and the objects. Each number is assigned to only one o~ject
used for identification,
and each object has only one number assigned to it. Common examples include Social Security
there is a strict one-to-one
numbers and numbers assigned to rugby players. In marketing research, nominal scales are used
correspondence between
the numbers and the for identifying respondents, brands, attributes, stores, and other o~jects.
objects When used for classification purposes, the nominally scaled numbers serve as labels for
classes or categories. For example, you might classify the control group as group l and the
experimental group as group 2. The classes are mutually exclusive and collectively exhaustive.
The objects in each class are viewed as equivalent with respect to the characteristic represented
by the nominal number. All objects in the same class have the same number and no two classes
have the same number. However, a nominal scale need not involve the assignment of numbers;
alphabets or symbols could be assigned as well. In the opening example, alphabets were assigned
to denote specific companies.
The numbers in a nominal scale do not reflect the amount of the characteristic possessed by the
o~jects. For example, a high Social Security number does not imply that the person is in some way
superior to those with lower Social Security numbers or vice versa. The same applies to numbers

FIGURE 8.1 Scale


An Illustration of

~ i ~
Nominal Numbers Finish
Primary Scales of
Assigned
Measurement
to Runners

~ ~ ~
Ordinal Rank Order Finish
of Winners

Third Second First


Place Place Place

Interval Performance
Rating on a 8.2 9.1 () 6
0-to- IO Scale

Ratio Time to Finish,


15.2 14.1 13.4
in Seconds
CHAPTER 8 • MEASUREMENT AND SCALING: FUNDAMENTALS AND COMPARATIVE SCALING 285

Primary Scales of Measurement


Permissible Statistics
Basic Common Marketing
Scale Characteristics Examples Examples Descriptive Inferential
Nominal Numbers identify Social Security Brand numbers, Percentages, mode Chi-square,
and classify objects numbers, numbering store types, sex binomial test
of rugby players classification
Ordinal Numbers indicate Lhe relative Quality rankings, Preference Percentile, median Rank-order
positions of the objects but rankings of teams rankings, market correlation,
not the magnitude of in a tournament position, social Friedman ANOVA
differences between them class
Interval Differences between objects Temperature Attitudes, opinions, Range, mean, Product-moment
can be compared; zero (Fahrenheit, index numbers standard correlations, I-tests,
point is arbitrary centigrade) deviation ANOVA, regression,
factor analysis
Ratio Zero point is fixed; ratios Length, weight Age, income, Geometric mean, Coefficient of
of scale values can costs, sales, harmonic mean variation
be computed market shares

assigned to classes. The only permissible operation on the numbers in a nominal scale is counting.
Only a limited number of statistics, all of which are based on frequency counts, are permissible.
These include percentages, mode, chi-square, and binomial tests (see Chapter 15). It is not mean-
ingful to compute an average Social Security number, the average sex of the respondents in a
survey, or the number assigned to an average department store, as in the following example.

Project Research Nominal Scale


--~·------------------------------------
In the department store project, numbers I through t O were assigned to the IO stores considered in the study
(see Table 8.2). Thus store number 9 referred to Sears. This did not imply that Sears was in any way supe-
rior or inferior to Neiman Marcus, which was assigned the number 6. Any reassignment of the numbers,
such as transposing the numbers assigned to Sears and Neiman Marcus, would have no effect on the
numbering system, because the numerals did not reflect any characteristics of the stores. It is meaningful to

Illustration of Primary Scales of Measurement


1it'i!U$.. "',Vl;:1 18\441

Nominal Scale
&;lk/A
- Ordinal Scale Interval Scale Ratio Scale
Preference
Preference Ratings $ Spent Last
No. Store 1-7 11-17 3 Months
I. Nordstrom 7 79 5 15 0
2. Macy's 2 25 7 17 200
3. Kmart 8 82 4 14 0
4. Kohl's 3 30 6 16 100
5. JCPenney 1 IO 7 17 250
6. Neiman-Marcus 5 53 5 15 35
7. Marshalls 9 95 4 14 0
8. Saks Fifth Avenue 6 61 5 15 100
9. Sears 4 45 6 16 0
10. Wal-Mart IO 115 2 12 10
286 PART II • RESEARCH DESIGN FORMULATION

make statements such as "75 percent of the respondents patronized store 9 (Sears) within the last month."
Although the average of the assigned numbers is 5.5, it is not meaningful to state that the number of the
average store is 5.5. •

Ordinal Scale
ordinal scale An ordinal scale is a ranking scale in which numbers are assigned to objects to indicate the
A ranking scale in which relative extent to which the objects possess some characteristic. An ordinal scale allows you to
numbers are assigned to determine whether an object has more or less of a characteristic than some other object, but not
objects to indicate the how much more or less. Thus, an ordinal scale indicates relative position, not the magnitude of
relative extent to which the differences between the objects. The object ranked first has more of the characteristic as
some characteristic is
compared to the object ranked second, but whether the object ranked second is a close second or
possessed. Thus it is
a poor second is not known. The ordinal scales possess description and order characteristics but
possible to determine
whether an object has more
do not possess distance (or origin). Common examples of ordinal scales include quali.ty
or less of a characteristic rankings, rankings of teams in a tournament, socioeconomic class, and occupational status. In
than some other object. marketing research, ordinal scales are used to measure relative attitudes, opinions, perceptions,
and preferences. In the opening example, the rank order of the most admired companies repre-
sented an ordinal scale. Apple with a rank of 1, was America's most admired company.
Measurements of this type include "greater than" or "less than" judgments from the respondents.
In an ordinal scale, as in a nominal scale, equivalent objects receive the same rank. Any
series of numbers can be assigned that preserves the ordered relationships between the objects.
For example, ordinal scales can be transformed in any way as long as the basic ordering of the
objects is maintained. 4 In other words, any monotonic positive (order-preserving) transformation
of the scale is permissible, because the differences in numbers are void of any meaning other
than order (see the following example). For these reasons, in addition to the counting operation
allowable for nominal scale data, ordinal scales permit the use of statistics based on centiles. It is
meaningful to calculate percentile, quartile, median (Chapter 15). rank order correlation
(Chapter 17), or other summary statistics from ordinal data.

-·----·--·-- ·--·--·---·--·-------
Project Research Ordinal Scale

Table 8.2 gives a particular respondent's preference rankings. Respondents ranked IO department stores in
order of preference by assigning rank I to the most preferred store, rank 2 to the second most preferred
store, and so on. Note that JCPenney (ranked l) is preferred to Macy's (ranked 2), but how much it is pre-
ferred we do not know. Also, it is not necessary that we assign numbers from I to 10 to obtain a preference
ranking. The second ordinal scale, which assigns a number 10 to JCPenney, 25 to Macy's, 30 to Kohl's, and
so on, is an equivalent scale, as it was obtained by a monotonic positive transformation of the first scale.
The two scales result in the same ordering of the stores according to prefercuce. •

ACTIVE RESEARCH

United Flying for Customer Satisfaction


Visit www.united.com and search the Internet using a search engine as well as your library's online
databases to obtain information on customer satisfaction with airlines.
As the marketing director for United Airlines, what marketing strategies wou1d you formulate to
enhance satisfaction?
How would you use nominal and ordinal scales to measure customer satisfaction with the major
airlines such as United and its competitors?
interval scale
A scale in which the
numbers are used to rate
Interval Scale
objects such that
numerically equal distances In an interval scale, numerically equal distances on the scale represent equal values in the
on the scale represent equal characteristic being measured. An interval scale contains all the information of an ordinal scale,
distances in the but it also allows you to compare the differences between objects. The difference between any two
characteristic being scale values is identical to the difference between any other two adjacent values of an interval
measured. scale. There is a constant or equal interval between scale values. The difference hetween l and 2
CHAPTER 8 • MEASUREMENT AND SCALING: FUNDAMENTALS AND COMPARATIVE SCALING 287

is the same as the difference between 2 and 3, which is the same as the difference between 5 and 6.
The distance between descriptors is known. A common example in everyday life is a temperature
scale. In marketing research, attitudinal data obtained from rating scales are often treated as inter-
val data. In the opening example of the most admired companies, the ratings on all the criteria
attributes represented an interval scale. 5
In an interval scale, the location of the zero point is not fixed, i.e., these scales do not possess the
origin characteristic. Both the zero point and the units of measurement are arbitrary. Hence, any
positive linear transformation of the form y = a + bx will preserve the properties of the scale. Here,
x is the original scale value, y is the transfom1ed scale value, b is a positive constant, and a is any
constant. Therefore, two interval scales that rate objects A, B, C, and Das I, 2, 3, and 4, or as 22, 24,
26, and 28, are equivalent. Note that the latter scale can be derived from the fonner by using a 20
and b 2 in the transforming equation. Because the zero point is not fixed, it is not meaningful to
take ratios of scale values. As can be seen, the ratio of D to B values changes from 2: 1 to become 7:6
when the scale is transformed. Yet ratios of differences between scale values are permissible. In this
process, the constants a and bin the transfonning equation drop out in the computations. The ratio of
the difference between D and B to the difference between C and B is 2: 1 in both the scales.

Project Research Interval Scale

In Table 8.2, a respondent's preferences for the 10 stores are expressed on a 7-point rating scale. We can see that
although Sears received a preference rating of 6 and Wal-Mart a rating of 2, this does not mean that Sears is pre-
fen-ed three times as much as Wal-Mart. When the ratings are transfonned to an equivalent l 1-to-17 scale (next
column), the ratings for these stores become 16 and 12, and the ratio is no longer 3 to L In contrast, the ratios of
preference differences are identical on the two scales. The ratio of the preference difference between JCPenney
and Wal-J\,1art to the preference difference between Neiman Marcus and Wal-Mart is 5 to 3 on both the scales. •

Statistical techniques that may be used on interval scale data include all of those that can be
applied lo nominal and ordinal data. In addition, you can calculate the arithmetic mean, standard
deviation (Chapter 15), product moment correlations (Chapter 17), and other statistics com-
monly used in marketing research. However, certain specialized statistics such as geometric
mean, harmonic mean, and coefficient of variation are not meaningful on interval scale data.
As a further iltustration, Federation Internationale de Football Association (FIFA) uses
ordinal and interval scaling to rank football teams of various countries.

Real Research Scaling the Football World

According to Federation Internationale de Football Association's (FIFA) (www.fifa.com) September 2008


men ·s rankings, Spain was at the top with 1,565 points and Italy was in the second spot with 1,339 points.
The top 10 countries in men's football are shown.

Ranking as of September 2008

ID Team Rank Points


A Spain 1,565
B Italy 2 1,339
C Germany 3 1,329
D Netherlands 4 1,295
E Croatia 5 1,266
F Brazil 6 1,252
G Argentina 7 1,230
H Czech Republic 8 1,134
I Portugal 9 1,120
J Turkey 10 1,033

The alphabets assigned to countries constitute a nominal scale, and the rankings represent an ordinal scale,
whereas the points awarded denote an interval scale. Thus country G refers to Argentina, which was ranked 7
288 PART II • RESEARCH DESIGN FORMULATION

and received 1,230 points. Note that the alphabets assigned to denote the countries simply serve the purpose of
identification and are not in any way related to their foothall-playing capabilities. Such information can be
obtained only by looking at the ranks. Thus, Croatia, ranked 5, played better than Turkey, ranked I 0. The lower
the rank, the better the perfo1mance. TI1e ranks do not give any infonnation on the magnitude of the differences
between countries, which can be obtained only by looking at the points. Based on the points awarded, it can be
seen that Italy, with 1,339 points, played only marginally better than Gennany, with 1,329 points. The points
help us to discern the magnitude of difference between countries receiving different ranks. 6 •

Ratio Scale
ratio scale A ratio scale possesses all the properties of the nominal, ordinal, and interval scales and, in addition,
The highest scale. It allows an absolute zero point. Thus, ratio scales possess the characteristic of origin (and distance, order, and
the researcher to identify or description). Thus, in ratio scales we can identify or classify objects, rank the o~jects, and compare
classify objects, rank-order intervals or differences. It is also meaningful to compute ratios of scale values. Not only is the
the objects, and compare difference between 2 and 5 the same as the difference between 14 and 17, but also 14 is seven times
intervals or differences. It is
as large as 2 in an absolute sense. Common examples of ratio scales include height, weight, age, and
also meaningful to compute
money. In marketing, sales, costs, market share, and number of customers are variables measured on
ratios of scale values.
a ratio scale. In the opening example, the rate of return of the most admired companies, not shown,
could be represented on a ratio scale.
Ratio scales allow only proportionate transformations of the form y = bx, where b is a positive
constant. One cannot add an arbitrary constant, as in the case of nn interval scale. An example of
this transformation is provided by the conversion of yards to feet (b = 3). The comparisons
between the objects are identical whether made in yards to feet.
All statistical techniques can be applied to ratio data. These include specialized statistics
such as geometric mean, harmonic mean, and coefficient of variation.
TI1e four primary scales (discussed here) do not exhaust the measurement-level categories. It is
possible to construct a nominal scale that provides partial information on order (the partially
ordered scale). Likewise, an ordinal scale can convey partial information on distance, as in the case
of an ordered metric scale. A discussion of these scales is beyond the scope of this text. 7

Project Research Ratio Scale

In the ratio scale illustrated in Table 8.2, a respondent is asked to indicate the dollar amounts spent in each
of the 10 stores during the last two months. Note that whereas this respondent spent $200 in Macy's and
only $JO in Wal-Mart, this person spent 20 times as much in Macy's as Wal-Mart. Also, the zero point is
fixed, because O means that the respondent did not spend anything at that store. Multiplying these numbers
by I00 to convert dollars to cents results in an equivalent scale.

Project Activities
In the context of the Sears project, develop questions to measure each of the following variables and iden-
tify the level of measurement in each question.

l. Marital status of respondent


2. Age of respondent
3. Annual household income before taxes
4. Familiarity with Sears
5. Importance of price in selecting a department store
6. Number of bank credit cards owned •

ACTIVE RESEARCH

Coaching Consumer Prefere11ces for Coach


Visit www.coach.com and search. the Internet using a search engine as well as your library's online
databases to obtain information on consumer preferences for leather goods.
How would you use interval and ratio scales tomeasure consumer preferences for leather goods?
As the marketing director for Coach, how would you use information on consumer preferences for
leather goods to increase your market share?
CHAPTER 8 • MEASUREMENT AND SCALING: FUNDAMENTALS AND COMPARATIVE SCALING 289

FIGURE 8.2
A Classification of
Scaling Techniques

1
. Comparative Noncomparative
Scales Scales

! !
!
Paired
I
Rank
1
Constant
1
Q-S.ort and
!
Continuous
1
Itemizec;l
Comparison Order. Sum Other Procedures Rating Scales , Rating Scales

l
! 1
Semantic
1
Likert Stapel
Differential

A Comparison of Scaling Techniques


The scaling techniques commonly employed in marketing research can be classified into
comparative scales comparative and noncomparative scales (see Figure 8.2). Comparative scales involve the direct
One of two types of scaling comparison of stimulus objects. For example, respondents might be asked whether they prefer
techniques in which there is Coke or Pepsi. Comparative scale data must be interpreted in relative terms and have only
direct comparison of ordinal or rank order properties. For this reason, comparative scaling is also referred to as
stimulus objects with one
nonmetric scaling. As shown in Figure 8.2, comparative scales include paired comparisons, rank
another.
order, constant sum scales, Q-sort, and other procedures.
The major benefit of comparative scaling is that small differences between stimulus objects
can be detected. As they compare the stimulus objects, respondents are forced to choose between
them. ln addition, respondents approach the rating task from the same known reference points.
Consequently, comparative scales are easily understood and can be applied easily. Other advan-
tages of these scales are that they involve fewer theoretical assumptions, and they also tend to
reduce halo or carryover effects from one judgment to another. The major disadvantages of
comparative scales include the ordinal nature of the data and the inability to generalize beyond the
stimulus objects scaled. For instance, to compare RC Cola to Coke and Pepsi, the researcher would
have to do a new study. These disadvantages are substantially overcome by the noncomparative
scaling techniques.
noncomparative scales Tu noncomparative scales, also referred to as monadic or metric scales, each object is
Orre of two types of scaling scaled iudependently of the others in the stimulus set. The resulting data are generally
techniques in which each assumed to be interval or ratio scaled. 8 For example, respondents may be asked to evaluate
stimulus object is scaled Coke on a l-to-6 preference scale ( 1 = not at all preferred, 6 = greatly preferred). Similar
independently of the other
evaluatious would be obtained for Pepsi and RC Cola. As can be seen in Figure 8.2, noncom-
objects in the stimulus set.
parative scales cau be contiuuous rating or itemized rating scales. The itemized rating scales
can be further classified as Likert, semantic differential, or Stapel scales. Noncomparative
paired comparison scaling is the most widely used scaling technique in marketing research. Given its importance,
scaling Chapter 9 is devoted to noncomparative scaling. The rest of this chapter focuses on compara-
A comparative scaling tive scaling techniques.
technique in which a
respondent is presented
with two objects at a time Comparative Scaling Techniques
and asked to select one
object in the pair according
Paired Comparison Scaling
to some criterion. The data As its name implies, in paired comparison scaling, a respondent is presented with two objects
obtained are ordinal in and asked to select one according to some criterion. The data obtained are ordinal in nature.
nature. Respondents may state that they shop in JCPenney more than in Sears, like Total cereal better
290 PART II • RESEARCH DESIGN FORMULATION

than Kellogg's Product 19, or like Crest more than Colgate. Paired comparison scales are
frequently used when the stimulus objects are physical products. Coca-Cola is reported to have
conducted more than 190,000 paired comparisons before introducing New Coke.9 Paired
comparison scaling is the most widely used comparative scaling technique.
Figure 8.3 shows paired comparison data obtained to assess a respondent's shampoo
preferences. As can be seen, this respondent made 10 comparisons to evaluate five brands. In
general, with n brands, [n(n - I )/2] paired comparisons are required to include all possible
pairings of objects. 10
Paired comparison data can be analyzed in several ways. 11 The researcher can calculate the
percentage of respondents who prefer one stimulus to another by summing the matrices of
Figure 8.3 for all the respondents, dividing the sum by the number of respondents, and multiply-
ing by I 00. Simultaneous evaluation of all the stimulus objects is also possible. Under the
assumption of transitivity, it is possible to convert paired comparison data to a rank order.
transitivity of Transitivity of preference implies that if brand A is preferred to B, and brand B is preferred to c·,
preference then brand A is preferred to C. To arrive at a rank order, the researcher determines the number of
An assumption made in times each brand is preferred by summing the column entries in Figure 8.3. Therefore, this
order to convert paired respondent's rank order of preference, from most to least preferred, is Head and Shoulders,
comparison data to rank
Jhirmack, Finesse, Pert, and Vidal Sassoon. It is also possible to derive an interval scale from
order data. It implies that if
paired comparison data using the Thurstone case V procedure. Refer to the appropriate literature
brand A is preferred to
for a discussion of this procedure. 12
brand B and brand B is
preferred to brand C, then
Several modifications of the paired comparison technique have been suggested. One
brand A is preferred to involves the inclusion of a neutral/no difference/no opinion response. Another extension is
brand C. graded paired comparisons. In this method, respondents are asked which brand in the pair is
preferred and how much it is preferred. The degree of preference may be expressed hy how
much more the respondent is willing to pay for the preferred brand. The resulting scale is a
dollar metric scale. Another modification of paired comparison scaling is widely used in obtain-
ing similarity judgments in multidimensional scaling (see Chapter 21).
Paired comparison scaling is useful when the number of brands is limited, because it
requires direct comparison and overt choice. However, with a large number of brands, the
number of comparisons becomes unwieldy. Other disadvantages are that violations of
the assumption of transitivity may occur, and order in which the objects are presented may bias
the results. Paired comparisons bear little resemblance to the marketplace situation that involves
selection from multiple alternatives. Also, respondents may prefer one object to certain others,
hut they may not like it in an absolute sense.

FIGURE 8.3
Obtaining Instructions
Shampoo We are going to present you with 10 pairs of shampoo brands. For each pair, please indicate
which one of the two brands of shampoo in the pair you would prefer for personal use.
Preferences Using
Paired Recording Form
Comparisons Jhirmack Finesse Vulal Sassoon Head & Shoulders Pert
Jhinnack 0 0 I 0
Finesse la 0 0
Vidal Sassoon I
Head & Shoulders 0 0 0 0
Pert 0
Number of
times prefrrredb 3 2 0 4
"A I in a particular box means that the brand in that column was preferred over the brand in the
con-esponding row. A O means that the row brand was preferred over the column brand.
bThe number of times a brand was prefen-ed is obtained by summing the l's in each column.
CHAPTER 8 • MEASUREMENT AND SCALING: FUNDAMENTALS AND COMPARATIVE SCALING 291

Real Research Paired Comparison Scaling

The most common method of taste testing is paired comparison. The consumer is asked to sample two
different products and select the one with the most appealing taste. The test is done in private, either in
homes or other predetermined sites. A minimum of 1,000 responses is considered an adequate sample.
Ocean Spray (www.oceanspray.com), North America's top producer of bottled and canned juices/juice
drinks, makes extensive use of taste tests in developing new products. Respondents are asked to sample new
drinks presented in pairs, evaluate them on taste and flavor, and choose the one they like more than the other.
laste tests showed that several consumers prefe1Ted white cranberries to the strong, tart taste of red cranber-
ries. Therefore, in early 2002, Ocean Spray added White Cranberry drinks, made with natural white cranberries
harvested a few weeks earlier than the red variety, and Juice Spritzers, lightly carbonated juice drinks, to its
product line in an effort to appeal to a broader range of consumers. After conducting taste tests, Ocean Spray
introduced in 2004 an innovative new product that combines the goodness offruitjuiee with the cool, refresh-
ing taste of tea that consumers crave when the weather heats up. When taste tests showed a clear preference over
competing energy drinks, in 2008 Ocean Spray introduced CranergyTM, a new juice drink that delivers natural
energy. Cranergy was available in two flavors--Cranberry Lift and Raspbeffy Cranberry Lift. 13 •

Experiential Which Is It? Coke or Pepsi


Research
Which cola drink would be the most popular among your friends in a taste test? Develop a script for testing
for preference between two cola drinks: Coke and Pepsi. Make sure the drinks are chilled to the same
temperature aud poured into three groups of identical cups. Each person will select a cup from the first
group. taste the cola, eat a cracker, and take a swallow of water. Repeat this when tasting from the second
and third groups. (The first group might be Pepsi, the second group Coke, and the third group Pepsi.) Afler
tasting from each of the three cola sample groups, have the respondents complete the survey form you
created. To avoid a biasing effect from the order of presentation of the samples, it will be important for half
of the respondents to have a rotated order of presentation: The first group Coke. the second group Pepsi, and
the third group Coke. Conduct the taste test on 30 respondents/students.

1. How many respondents correctly identified the two identical samples of cola?
2. Of those who conectly identified the two identical samples of cola, which cola was preferred: Coke
or Pepsi?
3. Of those who co1Tectly identified the two identical samples of cola, how many had no preference
between Coke and Pepsi?
4. In sum. who would you say is the winner in your taste test: Coke or Pepsi? Or, is it too close to tell? •

Rank Order Scaling


After paired comparisons, the most popular comparative scaling technique is rank order scaling.
rank order scaling In rank order scaling, respondents are presented with several objects simultaneously and asked to
A comparative scaling order or rank them according to some criterion. For example, respondents may be asked to rank
technique in which brands of toothpaste according to overall preference. As shown in Figure 8.4, these rankings are
re,sponderits are presented typically obtained by asking the respondents to assign a rank of l to the most preferred brand, 2 to
with several
the second most preferred, and so on, until a rank of n is assigned to the least preferred brand. Like
simultaneously and asked
paired comparison, this approach is also comparative in nature, and it is possible that the respon-
to order or rank them
~rrmrl,nn to sorne criterion.
dent may dislike the brand ranked l in au absolute sense. Furthermore, rank order scaling also
results in ordinal data. See Table 8.2, which uses rank order scaling to derive an ordinal scale.
Rank order scaling is commonly used to measure preferences for brands as well as attributes.
Rank order data are frequently obtained from respondents in conjoint analysis (see Chapter 21),
because rank order scaling forces the respondent to discriminate among the stimulus objects.
Moreover, as compared to paired comparisons, this type of scaling process more closely
resembles the shopping environment. lt also takes less time and eliminates intransitive responses.
If there are n stimulus objects, only (n I) scaling decisions need be made in rank order scaling.
However, in paired comparison scaling, [n(n - 1)/2] decisions would be required. Another
advantage is that most respondents easily understand the instructions for ranking. The major
disadvantage is that this technique produces only ordinal data.
292 PART II • RESEARCH DESIGN FORMULATION

FIGURE 8.4
Instructions
Preference for
Rank the various brands of toothpaste in order of preference. Begin by picking out the one
Toothpaste Brands
brand that you like most and assign it a number I. Then find the second mos! prefen-ed brand
Using Rank Order and assign it a number 2. Continue this procedure until you have ranked all the brands of
Scaling toothpaste in order of preference. The least prefen-ed brand should be assigned a rank of 10.
No two brands should receive the same rank number.
The criterion of preference is entirely up to you. There is no right or wrong answer. Just try
to be consistent.

Brand Rank Order


I. Crest
2. Colgate
3. Aim ---
4. Gleem
5. Sensodyne
6. Ultra Brite ----
7. Close Up
8. Pepsodent
9. Plus White
10. Stripe ·----

Finally, under the assumption of transitivity, rank order data can be conve1ted to equivalent
paired comparison data, and vice versa. Figure 8.3 illustrated this point. lt is possible to derive an
interval scale from rankings using the Thurstone case V procedure. Other approaches for deriving
interval scales from rankings have also been suggested. 14 The following example shows how rank
order scaling is used to determine the world's top brands.

Real Research World's Best-Known Brands

To be a strong competitor in today's American marketplace, companies rely heavily on brand recognition.
When consumers recognize a company or a product as a household name, preferably for good reasons
rather than as a result of bad publicity, brand equity is increased. lnterbrand (www.interbrand.com) is a
company dedicated to identifying, building, and expressing the right idea for a brand. Business
Week/Interbrand publish an annual ranking and value of the world's best-known brands. Here are the top
five for 2008: 15 •

The World's Five Most Valuable Brands

2008 Brand Value


Brand Rank {$
Coca-Cola 66.667
IBM 2 59.031
Microsoft 3 59.007
GE 4 53.086
Nokia 5 35.942
constant sum scaling
A comparative scaling
technique in which Another example of rank order scaling was America's most admired companies given in the
respondents are required to "Overview" section.
allocate a constant sum of
units such as points, dollars,
chits, stickers, or chips
Constant Sum Scaling
among a set of stimulus In constant sum scaling, respondents allocate a constant sum of units, such as points, dollars, or
objects with respect to chips, among a set of stimulus objects with respect to some criterion. As shown in Figure 8.5,
some criterion. respondents may be asked to allocate 100 points to attributes of a toilet soap in a way that reflects
CHAPTER 8 • MEASUREMENT AND SCALING: FUNDAMENTALS AND COMPARATIVE SCALING 293

FIGURE 8.5
Importance of I11structio11s
Toilet Soap Below are eight attributes of toilet soaps. Please allocate 100 points among the attributes so
that your allocation reflects the relative importance y()n attach to each attribute. The more
Attributes Using points an attribute receives, the more important the attribute is. If an attribute is not at all
a Constant Sum important, assign it zero points. If an attribute is twice as important as some other attribute,
Scale it should receive twice as many points.
~flmn
Average Responses of 11iree Segments
Attribute Segment I Segment JI Segment III
l. Mildness 8 2 4
2. Lather 2 4 17
3. Shrink.age 3 9 7
4. Price 53 17 9
5. Fragrance 9 0 19
6. Pack.aging 7 5 9
7. Moisturizing 5 3 20
8. Cleaning power 13 60 15
Sum 100 100 100

the importance they attach to each attribute. If an attribute is unimportant, the respondent assigns
it zero points. If an attribute is twice as important as some other attribute, it receives twice as
many points. The sum of all the points is l 00. Hence, the name of the scale.
The attributes are scaled by counting the points assigned to each one by all the respondents
and dividing by the number of respondents. These results are presented for three groups, or
segments, of respondents in Figure 8.5. Segment I attaches overwhelming importance to price.
Segment II considers basic cleaning power to be of prime importance. Segment III values
lather, fragrance, moisturizing, and cleaning power. Such information cannot be obtained from
rank order data unless they are transformed into interval data. Note that the constant sum also
has an absolute zero-10 points are twice as many as 5 points, and the difference between 5 and
2 points is the same as the difference between 57 and 54 points. For this reason, constant sum
scale data are sometimes treated as metric. Although this may be appropriate in the limited
context of the stimuli scaled, these results are not generalizable to other stimuli not included in
the study. Hence, strictly speaking, the constant sum should be considered an ordinal scale
because of its comparative nature and the resulting lack of generalizability. It can be seen
that the allocation of points in Figure 8.5 is influenced by the specific attributes included in the
evaluation task.
The main advantage of the constant sum scale is that it allows for fine discrimination
among stimulus objects without requiring too much time. However, it has two primary dis-
advantages. Respondents may allocate more or fewer units than those specified. For example, a
respondent may allocate 108 or 94 points. The researcher must modify such data in some way
or eliminate this respondent from analysis. Another potential problem is rounding error if too
few units (e.g., points) are used. On the other hand, the use of a large number of units may be
too taxing on the respondent and cause confusion and fatigue.

ACTIVE RESEARCH

Lexus: Luminescence of Luxury


Visit www.lexus.com and search the Internet using a search engine as well as your library's online data-
bases to obtain information on consumer purchase intentions for luxury cars.
As the marketing manager for Lexus, how would you use information on consumer purchase inten-
tions for luxury cars to increase your sales?
Would you use a comparative scaling technique to measure consumer purchase intentions for luxury
cars? If yes, which technique?
294 PART 11 • RESEARCH DESIGN FORMULATION

Q-Sort and Other Procedures


Q-sort scaling Q-sort scaling was developed to discriminate quickly among a relatively large number of
A comparative scaling objects. This technique uses a rank order procedure in which objects are sorted into piles based
technique that uses a rank on similarity with respect to some criterion. For example, respondents are given I00 attitude
order procedure to sort statements on individual cards and asked to place them into 11 piles, ranging from "most highly
objects based on similarity agreed with" to "least highly agreed with." The number of objects to be sorted should not be Jess
with respect to some
than 60 nor more than 140; 60 to 90 objects is a reasonable range. The number of objects to be
criterion.
placed in each pile is prespecified, often to result in a roughly normal distribution of objects over
the whole set.
Another comparative scaling technique is magnitude estimation. 16 In this technique,
numbers are assigned to objects such that ratios between the assigned numbers reflect ratios on
the specified criterion. For example, respondents may he asked to indicate whether they agree
or disagree with each of a series of statements measuring attitude toward department stores..
Then they assign a number between Oto 100 to each statement to indicate the intensity of their
agreement or disagreement. Providing this type of number imposes a cognitive burden on the
respondents. Finally, mention must be made of Guttman scaling, or scalogram analysis, which
is a procedure for determining whether a set of objects can be ordered into an internally consis-
tent, unidimensional scale.

International Marketing Research


In the four primary scales, the level of measurement increases from nominal to ordinal to
interval to ratio scale. This increase in measurement level is obtained at the cost of complex-
ity. From the viewpoint of the respondents, nominal scales are the simplest to use, whereas
the ratio scales are the most complex. Respondents in many developed countries, due to
higher education and consumer sophistication levels, are quite used to providing responses
on interval and ratio scales. However, it has been argued that opinion formation may not be
well crystallized in some developing countries. Hence, these respondents experience diffi-
culty in expressing the gradation required by interval and ratio :,Cales. Preferences can,
therefore, be best measured by using ordinal scales. In particular, the use of binary scales
(e.g., preferred/not preferred), the simplest type of ordinal scale, has been recommended. 17
For example, when measuring preferences for jeans in the United States, Levi Strauss & Co.
could ask consumers to rate their preferences for wearing jeans on specified occasions using
a 7-point interval scale. However, consumers in Papua New Guinea could be shown a pair of
jeans and simply asked whether or not they would prefer to wear it for a specific occasion
(e.g., when shopping, working, relaxing on a holiday, etc.). The advantage of selecting the
primary scales to match the profile of the target respondents is well illustrated by the
Japanese survey of automobile preferences in Europe.

Real Research Car War-Japan Making a Spearhead


For the first time, European journalists had given their car-of-the-year award to a Japanese model-
Nissan's new British-made Miera, a $10,000 subcompact. This came as a big bluw to the European
automakers that have been trying to keep the Japanese onslaught at bay. "They will change the compet-
itive balance," warns Bruce Blythe, Ford of Europe Tnc.'s head of business strategy. How did the
Japanese do it'!
Nissan conducted a survey of European consumers' preferences for automobiles using interval
scales to capture the magnitude of preference differences. The use of interval scales enabled Nissan
to compare the differences between automobile features and determine which features were pre-
ferred. The findings revealed distinct consumer preferences. So the Japanese made inroads by trans-
planting their production and building technical centers in Europe to customize their cars to local
styling tastes and preferences. Nissan introduced new models in Europe in 2009 in hopes of lifting
recent sagging sales in that market. The European automakers need to be on guard against such fierce
competition. 18 •

Potrebbero piacerti anche