The Marketing Book 5th Edition

188 The Marketing Book

Rank order scales

Respondents are asked to rank order a list of
objects/items against a stated criterion, e.g.
taste, power etc. Rank order scales are ordinal,
thus respondents are only able to show the order
of their preferences; the research cannot infer
anything about the ‘distances’ between the
items, i.e. it is not possible to say by how much
the first item is preferred to the second etc.

Constant sum rating scales

This method overcomes the drawbacks of rank
order scales. The respondent is allocated a
constant sum (they may be expressed in cur-
rency or some other units), usually a round
number, 100 units etc., and asked to allocate
them between the given items in a way which
reflects the object’s attributes under investiga-
tion. This not only shows the rank order of the
items, but also the size of the preference
distances.
Attitude scales, which combine many rat-
ing scales, are an attempt to overcome the
unrepresentativeness that may arise from infer-
ring an individual’s overall response to some
object etc. by measuring their attitude to only a
single aspect of that object etc.; attitude scales
try to measure several facets of an individual’s
attitude to an object, person etc.

Likert or summated scales

Devised in 1932, the Likert or summated scales
require that respondents indicate their degree
of agreement or disagreement with a number of
statements concerning the attitude being meas-
ured. Their responses are given a numerical
value and/or sign to reflect the strength and
direction of the respondent’s reaction to the
statement. Thus, statements with which the
respondent agrees are given positive or high
values, while those with which they disagree
are given negative or low marks. Scales may
run, for example, from 1 to 5, from 5 to 1 or
from +2, via zero, to –2. Statements should give
the respondent the opportunity to express clear,

unambiguous statements, rather then neutral,

ambiguous ones.

Semantic differential scales

These are arguably the most widely used of all

attitude scales. Respondents show the position

of their attitude to the research object on a

seven-point itemized scale, thus revealing both

the strength and direction of the attitude. The

extremities of the continuum are ‘anchored’ by

a pair of polarized adjectives or adjectival

statements.

For example, respondents are asked to

record their attitude towards a certain law

firm:

Unfriendly ............Friendly

Modern ............Old fashioned

Efficient ............Inefficient

Slow ............Fast

Pleasant ............Unpleasant

Osgoodet al. (1957), who devised the scale,

developed some 50 pairs of bipolar adjectives

grouped to measure three fundamental compo-

nents of attitude:

1 Evaluative– negative/positive, good/bad.

2 Activity– active/passive, fast/slow.

3 Potency– weak/strong.

If phrases rather than words are used, then the

scale will have more meaning for respondents

(Dickson and Albaum, 1977). Luck and Rubin

(1987) recommend that no side of the scale

should be exclusively reserved for either the

positive or the negative aspect of the pairs, as

this tends to allow respondents to tick only

down one side – the ‘halo’ effect.

Semantic differential scales may be ana-

lysed in two main ways:

 Aggregate analysis– where the score is

summed for each respondent for all pairs of

words/statements, resulting in a numerical

value of their attitude. Individual aggregate