Você está na página 1de 8

Chapter 6-- Measurement and Scaling

Chapter Summary: The data used in a research consists of quantitative variables like price,
income, sales etc., and qualitative variables like knowledge, performance, character etc. The
qualitative information must be converted into numerical form for further analysis. This is
possible through measurement and scaling techniques. A common feature of survey based
research is to have respondent’s feelings, attitudes, opinions, etc. in some measurable form. This
process of conversion of non quantitative information into quantitative information required
scaling techniques.In this chapter the students will learn about--- Measurement and Scaling,
Issues in Attitude Measurement, Levels of Measurement Scales, Types of Scaling Techniques,
Comparative Scales, Non- comparative Scales and Selection of an Appropriate Scaling
Technique.

Introduction
Measurement is an integral part of the research process. It is concerned with the assignment of
numbers to the given variables. Marketing research is concerned with a number of attitudinal
measures. The general meaning of the term attitude is a settled behavior that is shown by a
person in a given situation. This behavior is not measurable but the researchers have derived
certain methods and scales to conveniently measure the attitudes of people in order to predict the
consumer behavior of the target market.

Scales of Measurement and Types of Scales


There are four types of attitude measurement scales. These are the :-

1. Nominal Scales 2. Ordinal Scales 3. Interval Scales 4. Ratio Scales

Nominal Scales:- The lowest level is the nominal scale. This may be thought of as the “naming”
level. For example, when we ask subjects to name their marital status, they will respond with
words—not numbers—that describe their status such as “married,” “single,” “divorced,” etc.
Notice that nominal data do not put subjects in any particular order. There is no logical basis for
saying that one category such as “single” is higher or lower than any other.

The next level is ordinal. At this level we put subjects in order from high to low. For instance, an
employer might rank order applicants for a job on their professional appearance. Traditionally,
we give a rank of 1 to the subject who is highest, 2 to the next highest, and so on. It is important
to note that ranks do not tell us by how much subjects differ. If we are told that Julie has a rank
of 1 and Ram has a rank of 2, we do not know if Juli’s appearance is greatly superior to Ram’s or
only slightly superior. To measure the amount of difference among subjects, we use the next
levels of measurement.

K.Shivam Page 1
Measurements at the interval and ratio levels have equal distances among the scores they yield.
For example, when we say that Jadav weighs 120 Kg and Shyam weighs 130 Kg, we know by
how much the two subjects differ. Also, note that a 10 Kg difference represents the same amount
regardless of where we are on the scale. For instance, the difference between 120 and 130 Kg is
the same as the difference between 220 and 230 Kg.

The ratio scale is at a higher level than the interval scale because the ratio has an absolute zero
point that we know how to measure. Thus, weight is an example of the ratio scale because it has
an absolute zero that we can measure.
The interval scale, while having equal intervals like the ratio scale, does not have an absolute
zero. The most common examples of interval scales are scores obtained using objective tests
such as multiple-choice tests of achievement. It is widely assumed that each multiple-choice test
item measures a single point’s worth of the trait being measured and that all points are equal to
all other points—making it an interval scale (just as all Kg are equal to all other Kg of weight).
However, such tests do not measure at the ratio level because the zero on such tests is arbitrary—
not absolute. To see this, consider someone who gets a zero on a multiple-choice final
examination. Does the zero mean that the student has absolutely no knowledge of or skills in the
subject area? Probably not. He or she probably has some knowledge of simple facts, definitions,
and concepts, but the test was not designed to measure at the skill level at which the student is
operating. Thus, a score of zero only indicates that they know nothing on that test — not that
they have zero knowledge of the content domain.
Here’s a summary of the levels:
__________________________________________________________
Lowest
Level Scale Characteristic
__________________________________________________________
Nominal naming

Ordinal ordering

Interval equal interval without absolute zero

Ratio equal interval with absolute zero

__________________________________________________________
Highest
__________________________________________________________

K.Shivam Page 2
Rating Scale
In a rating scale the rater evaluates a person or an object along a scale by assigning a numeric
value. The various types of rating scales used in marketing research are as under:
a. Graphic rating Scale: The interviewee indicates how he feels on a graphic rating scale.
A general graphic rating scale is known as the thermometer chart which is presented as
under.
100------------------- Very Good
50-------------------- Indifferent
0--------------------- Very Bad

On a scale of 0-100 how would you rate the book that you just read.
b. Itemized rating Scale: The interviewee is asked to select one of a restricted number of
categories. The number generally varies between two and eleven points. The itemized
scale shown below.
Are you interested in buying an LED Television?
Very Much Somewhat Not at all

c. Comparative rating scale: Interviewers judge a person, object or other phenomenon


against some other person, object or other phenomenon. This scale can take many forms
one of them being rank order scale:-

Assign 1 to the most desirable, 2 to next and so on……………………………………………

Philips Sony Samsung LG

d. Paired Comparison Rating Scale: The interviewer makes a number of comparisons in


which they select among objects two at a time. For example

Evaluate between the following refrigerator brands

LG and Kelvinator Wirlpool and Sharp

K.Shivam Page 3
Attitude Scales

These scales are those that measure a person’s predisposition towards another person object or
event. These scales differ from rating scales in that they are more complex and consist of multi
item rating scales.

a. Likert Scales: Likert scaling is a bipolar scaling method, measuring either positive or
negative response to a statement.

(Traditionally a five-point scale is used.)

A large Television is a good for the eyes:

1. Strongly disagree
2. Disagree
3. Neither agree nor disagree
4. Agree
5. Strongly agree

b. Semantic Differential Scale: The semantic differential scale asks a person to rate a
product, brand, or company based upon a seven-point rating scale that has two bipolar
adjectives at each end. The following is an example of a semantic differential scale
question.

Example: Would you say our web site is:

(7) Very Attractive


(6)
(5)
(4)
(3)
(2)
(1) Very Unattractive

K.Shivam Page 4
Notice that unlike the rating scale, the semantic differential scale does not have a neutral or
middle selection. A person must choose, to a certain extent, one or the other adjective.

c. Dichotomous Scale: The dichotomous question is generally a "yes/no" question. An


example of the dichotomous question is: Have you ever purchased a product or service
from our website?

1. Yes

2. No

Validity Assessment
Validity and reliability are two concepts frequently associated with both research theory and
marketing research as practiced in the real world.

Validity: May be defined as a measuring instrument is valid when it measures what


it is supposed to measure. The instrument is valid to the extent that its
measurements are free from systematic error (bias).

Reliability: May be defined as a measuring instrument is reliable when the results it


delivers are consistent. The instrument is reliable to the extent that its
measurements are free from nonsystematic (random) error.

The major approaches to the assessment of validity are the following:

1. Content Validity: Content validity, sometimes referred to as face validity, concerns the extent
to which the measurement instrument “appears to be” measuring the characteristics it is
intended to measure. Its assessment is necessarily subjective and usually involves the
judgment of experts. For example, consider a questionnaire designed to measure students’
attitudes regarding the university they attend. If the questionnaire has omitted any reference
to such items as the quality of instruction, the availability of extracurricular recreation
activities, etc., one would suspect that the instrument lacks content validity.
2. Construct Validity: Construct validity deals with the degree to which the measurement scale
represents and acts like the concept being measured. In general, construct validity is present
whenever our measure of a particular concept is related to measures of other related concepts
in a theoretically expected way. For example, we might expect “attitude towards malpractice
during exam.” If subsequent measurement finds that one’s attitude has little to do with the
frequency of malpractice during exam, we will not be able to assume that our research has

K.Shivam Page 5
achieved construct validity. There are two major approaches commonly used in evaluating
construct validity. They are explained in the following sections.
(a) Convergent validity It is defined as the degree of association between two maximally
different measures which purport to measure essentially the same concept. For
example, suppose we identify the construct of “readiness to go to college” and we
develop a scale to measure it. We could determine convergent validity by correlating
our instrument with one that was designed to measure, say, “readiness to learn.”
Intuitively, these two constructs should display a high degree of association. If they
do, they can be said to exhibit convergent validity and thus, some degree of construct
validity.
(b) Discriminant validity It is largely the opposite of convergent validity in that it can be
defined as the degree to which our measurement scale may be differentiated from
other scales purporting to measure maximally different concepts. For example, if we
correlated our “readiness to go to college” scale with, say, “readiness watch a movie”,
we would expect to find a high degree of negative correlation. If we did not find this,
it would suggest that our measurement device is really not measuring what it is
supposed to be measuring.
3. Criterion-related Validity: This form of validity is concerned with the degree to which the
scale under study is able to predict a variable which is designated a criterion. There are two
general subtypes which are examined in the following sections.
(a) Prediction validity It is the extent to which a future level of some criterion variable
can be predicted by current measurement on the scale of interest. Here the emphasis
rests primarily in the criterion (Predicted) variable rather than the measured variable.
For example, the usage of our scale of “readiness to go to college” to predict, the
number of days of classes taken by an individual in the next six months is an
assessment of predictive validity.
(b) Concurrent validity Concurrent validity, like predictive validity, is concerned with
comparing our measurement with some external criterion of success. These two
approaches differ primarily from a time perspective. Whereas assessment of
predictive validity involves comparing a predicted value associated with the measured
characteristics with the future value which actually occurs, determination of
concurrent validity involves the comparison of our measurement with others
measures that have been made at about the same point in time. For example, we may
wish to compare our own results with those obtained by a well-regarded professional
polling organization in order to obtain an indication of the concurrent validity of our
measurements.

K.Shivam Page 6
Reliability Assessment
Reliability refers to the consistency and stability of a score from a measurement scale. It is
differentiated from validity in that the former is concerned with the consistency issue while the
latter is concerned with the accuracy issue. A reliable instrument is one that will derive
consistent results and be relatively free from random error. These are three major approaches to
the assessment of reliability, viz.,

1. Stable Approach The stability approach to the assessment of reliability involves application
of the measuring instrument to the same people or objects at two points in time.
(a) Test-retest em method If the instrument, is reliable, and if the individuals or objects
have not changed during the time between measurements, the first measurements, for
each individual or object should correspond very closely to the second. This method
is known as test-retest method. The test-retest assessment of reliability is not
appropriate whenever there is reason to believe that the first measurement will have
an effect on the second, in which case the second measurement will include factors
other than the characteristic presumed to be measured. However, if the observations
from the two separate measurements to exhibit consistency, we will have support for
the reliability of our measurement device.
2. Internal Consistency Method This method assesses the homogeneity of a set of items. The
basic rationale for these types of reliability assessments rests on the fact that items in a scale
should behave similarly. The specific techniques included in this method are the split-half
similarly. The specific techniques included in this method are the split-half technique.
Cronback-Alpha, and Kuder and Richardson’s KR-20 formula.
(a) Split-Half technique It is accomplished by splitting a multi-item scale in half, and
then correlating the summated results of the scores in the first half with those in the
second half. Usually the items in the scale are randomly assigned to one half or the
other. However, the major problem with this estimate of reliability is that the estimate
of the coefficient of reliability is wholly dependent upon the manner in which the
items were split. As a result, it is difficult to interpret this coefficient as the best
estimate.
(b) Cronback-Alpha An internal consistency technique which overcomes the above
mentioned criticism is called Cronback-Alpha. In essence, this technique computes
the mean reliability coefficient estimates for all possible ways of splitting a set of
items in half. This results in a better estimate of reliability.
3. Equivalence Approach In this approach we design two measurement devices with the
intention of having two instruments that are “equivalent”. If, when applied to the same
persons or objects, these two instruments deliver comparable results, the reliability of each is
supported. For example, we may construct two different list of attitude toward Titan watches.
If there is high correlation between the scores recorded by the two instruments the reliability
of each supported. The major problem of this approach is the designing of forms that are

K.Shivam Page 7
truly equivalent, as the absence of similar results could result from either a lack of reliability
or a lack of equivalence.

Review Questions
1. What is the more common name given to ordinal scales?

2. Why would a marketing researcher employ a dollar metric scale?

3. A researcher wants to measure consumer preference between 9 brands of vegetable oil and has
decided to use the paired comparison method. How many pairs of brands will the researcher
present to the respondents?

4. Explain what is meant by a semantic differential scale.

5. Discuss briefly the issues involved in attitude measurement.


6. Differentiate between ranking scales and rating scales. Which one of these
scales is better for measuring attitudes?
7. In what type of situation is the Q-sort technique more appropriate?
8. Name any four situations in commerce where you can use the Likert scale.
9. Construct a Rank Order Scale to measure toothpaste preferences. Discuss its
advantages and disadvantages.
10. Construct a Semantic differential scale to measure the experiences of
respondents in using Brand-X of shaving cream (assume that all the
respondents use that brand).

K.Shivam Page 8

Você também pode gostar