Research Methods for Business

NATURE OF MEASUREMENT
 Measurement – a part of research and an aspect of
research design
Contents…
 Measurement consists of assigning numbers to
1. Nature of Measurement empirical events, objects or properties, or activities in
2. Four basic measurement scales compliance with a set of rules.
3. Measurement errors
4. Measurement of attitude
5. Good measurement

Bui Huy Hai Bich 03/11/2022 1 Bui Huy Hai Bich 03/11/2022 2
Research Methods for Business

 Operationalizing = operationally defining a concept to render

• Concepts and constructs are used at theoretical levels; it measurable
variables are used at the empirical level.
- Behavioral Observable An index of
• Variables accept numerals or values for the purpose of dimensions and measurement
testing and measurement. - Facets measurable of the
- Properties elements concept
• 2 types of variables:
- Objective and precise measurement (ex: the length and  An operational definition does not describe the correlates of
width of a room/ table…) the concept.
- Subjective and nebulous or not clear (ex: satisfaction, ◦ Ex: success in performance cannot be a dimension of
achievement motivation
loyalty, thirst…)
 Operationally defining a concept does not consist of
• Concepts, constructs, and variables may be defined delineating the reasons, antecedents, consequences, or
descriptively or operationally. correlates of the concept

Bui Huy Hai Bich 03/11/2022 3 Bui Huy Hai Bich 03/11/2022 4
Research Methods for Business

Construct
Reward Employee
positive impact
LEARNING System performance
EFFECTIVENESS
Dimension Dimension
Theoretical World
Retention
Understanding (recall) Application

Solve Integrate Empirical World

Answer Give Recall material
problems with other
questions appropriate after some
applying relevant Turn-over rate,
correctly examples lapse of time
concepts material Salary
Bonus linear correlation Absentism,
Element Element Element Element Element
Nonmonetary award Products / hours

Bui Huy Hai Bich 03/11/2022 5 Bui Huy Hai Bich 03/11/2022 6
Research Methods for Business

FOUR BASIC MEASUREMENT SCALES Ordinal scale

 Categorizes the variables
 Most simple scale.  Rank-order the categories in some meaningful way to
reflect higher–lower level of an attribute.
 Numerical values are assigned to different categories or
groups of subjects  Ordinal scales = nominal scale + an indication of order.
 However, we do not know the magnitude of differences
 Requirement: + mutually exclusive (nonoverlapping)
 The research results will be unchanged regardless the
+ collectively exhaustive (no category number series being used to represent the rank
into which respondents would normally fall)  Example: consumer preferences for Mobile phone brands

 Example: marital status, gender, company’s industry.

Bui Huy Hai Bich 03/11/2022 7 Bui Huy Hai Bich 03/11/2022 8
Research Methods for Business

Interval scale Ratio scale

 Use of numbers to rank objects or events such that the  Interval scale + “zero value”.
differences between the numerals correspond to the
differences between the objects/events on the properties  Zero value represents the state of “non-existence” of
the property of interest.
being measured.
 Ratio data represent the actual amounts of a variable.
 Zero is just a subjective assigned value

 Interval scale = Ordinal scale + “equal differences”  Example: Measures of physical dimensions such as
weight, height, distance.
 Examples: Calendars, Celsius temperature scales
 In business: money values, population counts,
 Researchers treat many attitude scales as interval. distances, return rates, and amounts of time.

Bui Huy Hai Bich 03/11/2022 9 Bui Huy Hai Bich 03/11/2022 10
Research Methods for Business

Levels of Measurement Scale

Nature of Attitudes
An attitude is a learned, stable predisposition to respond to
Nominal Classification oneself, others, objects, or issues in a consistent way.
◦ Cognitively based: memories, evaluations & beliefs
Classification I think bread is healthier than instant noodle for breakfast
Ordinal
Order
◦ Affectively based: feelings, intuition, values & emotions.
Classification Distance I hate instant noodle.
Interval
Order
◦ Behaviorally based: expectations and intentions.
Classification Distance I intend to eat more bread for breakfast.
Ratio
Order Natural Origin
Attitudes reflect past experience & shape future behavior.

Bui Huy Hai Bich 03/11/2022 11 Bui Huy Hai Bich 03/11/2022 12
Research Methods for Business

Attitude Scaling
Selecting a measurement scale for attitude
 Scaling is the “procedure for the assignment of
numbers (or symbols) to a property of objects to apply Requires the consideration of 8 factors:
some characteristics of numbers to the property”.
 Attitude scaling means assessing an attitude using a
number that represents a person’s score on an a. Research Objectives
attitudinal continuum.
 To measure characteristics of the participants.
extremely favorable  extremely unfavorable
 To use participants as judges of the objects
 Example: your attitude toward your university could be
measured on numerous scales that capture indicators presented to them.
of the different dimensions of your awareness, feelings,
or behavioral intentions toward the school.

Bui Huy Hai Bich 03/11/2022 13 Bui Huy Hai Bich 03/11/2022 14
Research Methods for Business

b. Response types
(Rating/Ranking/Categorization/Sorting)
c. Data Properties
 A rating scale: to give a score to an object.
Ex: Participants are asked to evaluate the styling of a new automobile
on a 7-point scale.
 Nominal scales classify data into categories.
 Ordinal data show relationships of more than and
 Ranking scales: to set order among properties or objects.
Ex: Participants choose which one of a pair of cars has more attractive
less than but have no distance.
styling.
 Interval scales have both order and distance but no
 Categorization: to put into groups. unique origin.
Ex: Respondents select their gender.
 Ratio scales possess all four properties’ features
 Sorting: to sort cards into piles using researcher’s criteria .
Ex: The cards might contain photos or verbal statements such as
various descriptors of the car’s performance.

Bui Huy Hai Bich 03/11/2022 15 Bui Huy Hai Bich 03/11/2022 16
Research Methods for Business

e. Balanced or Unbalanced scale

d. Number of Dimensions
 Balanced: equal number of categories above and
Measurement scales are either uni-dimensional or below the midpoint.
multidimensional.
 Unbalanced: unequal number. Used when nearly all
 With a uni-dimensional scale, only one attribute of participants’ scores will lean in one direction.
the participant or object is measured.

 A multidimensional scale recognizes that an

object might be better described with several f. Forced or Unforced Choices scale
dimensions than on a uni-dimensional continuum.  Unforced-choice: allows “no opinion”/ “don’t know”.
 Forced-choice: participants must select one of
options.

Bui Huy Hai Bich 03/11/2022 17 Bui Huy Hai Bich 03/11/2022 18
Research Methods for Business

h. Respondent Errors
g. Number of Scale Points
 Tendencies to make errors of central tendency and
halo effect.
 As the number of points increases, reliability
increases.  Some respondents are reluctant to give extreme
judgments (central tendency).
 11-point Scale may produce more valid results
than 3-, 5-, or 7-point scales.  Respondents may also be “easy raters” or “hard
raters” (leniency error).
 A larger number of scale points are needed to
produce accuracy when using single-dimension  The halo effect is the systematic bias by carrying
over a generalized impression of the subject from
versus multiple-dimension scales. one rating to another.

Bui Huy Hai Bich 03/11/2022 19 Bui Huy Hai Bich 03/11/2022 20
Research Methods for Business

1. Dichotomous scale: to elicit a Yes/ No answer 3. Likert Scales: to examine how strongly
Ex: I plan to buy a Laptop in the 6 months. subjects agree/ disagree with statements on a 5-
[ ] yes [ ] no point scale
Ex: The university librarian is helpful.
2. Category scale: to elicit a single response from [ ] strongly agree
multiple items [ ] disagree
[ ] neither agree nor disagree
Ex: Which supermarket do you often visit
[ ] agree
[ ] BigC [ ] Coop Mart [ ] Maximark [ ] Other [ ] strongly agree.

Bui Huy Hai Bich 03/11/2022 21 Bui Huy Hai Bich 03/11/2022 22
Research Methods for Business

4. Summated rating scales (Itemized Rating 5. Semantic Differential Scales: bipolar adjectives
scale): used
 Statements expressing a favorable/unfavorable attitude. University librarian:
 Each response is given a score. The scores are then Skillful __:__:__:__:__:__:__ Unskillful
summed.
Beautiful __:__:__:__:__:__:__ Ugly
1 2 3 4 5
Neither 6. Numerical scales: similar to the semantic differential
Very unlikely Unlikely Unlikely Nor Likely Very Likely scale, with the difference that numbers on a 5-point scale
Likely
provided

Bui Huy Hai Bich 03/11/2022 23 Bui Huy Hai Bich 03/11/2022 24
Research Methods for Business

7. Stapel Scale: simultaneously measures both the 8. Constant-Sum Scales: distribute a given number
direction and intensity of the attitude toward the items of points across various items

9. Graphic Rating Scales: place a mark at the

appropriate point on the line to indicate the answer

Bui Huy Hai Bich 03/11/2022 25 Bui Huy Hai Bich 03/11/2022 26
Research Methods for Business

Forms of Ranking Scales

 Paired-comparison scale:

 To tap preferences between two or among more For each pair of beer brand, place a check beside the one you
would prefer if you had to choose between the two
objects or items ____ Tiger _____ Tiger _____ Saigon
◦ Paired comparisons: among a small number of ____ Saigon _____ Heineken _____ Heineken

objects, choose between two objects at a time.

 Forced ranking scale:
◦ Forced choice: rank objects relative to one another,
Rank the following beer brands in your order of preference.
among the alternatives provided. Place “1” next to your most preferred, “2” by the second most
choice, and so forth.
___ Tiger
___ Heineken
___ Saigon
___ Zorok

Bui Huy Hai Bich 03/11/2022 27 Bui Huy Hai Bich 03/11/2022 28
Research Methods for Business

• Characteristic:
status, Ethnic
Three major criteria:
• Presence of other
group, Social person.
class  Validity is the extent to which a test measures what
• Feelings may be
• Temporary Respondent Situation hidden if we actually wish to measure.
factors: anonymity is not
fatigue, ensured
boredom,  Reliability has to do with the accuracy and precision
anxiety, etc.
Sources of Errors A defective instrument
can cause distortion:
of a measurement procedure.
• Rewording,
paraphrasing, or + too confusing and
reordering questions.Measurer Instrument ambiguous.
 Practicality is concerned with a wide range of
• Appearance and action + poor selection of factors of economy, convenience, and
introduce bias. interpretability.
content items
• Unconscious
prompting: smiles,
nods, etc..
Bui Huy Hai Bich 03/11/2022 29 Bui Huy Hai Bich 03/11/2022 30
Research Methods for Business

Stability

 Consistent results with repeated attempts on the same

 A scale is reliable if it provides consistent results. person with the same instrument.
 A necessary, but not a sufficient condition of validity.  This leads to a test-retest arrangement, with
 Reliability  error free – without bias comparisons between the two tests to learn how
reliable they are.
 Reliable scales:
◦ Ex: a questionnaire measuring a concept is administered to a
 stability and internal consistency. set of respondents NOW and 6 months later.

Bui Huy Hai Bich 03/11/2022 31 Bui Huy Hai Bich 03/11/2022 32
Research Methods for Business

Internal Consistency
 Uses only one administration of an instrument to assess
the internal consistency or homogeneity among the  External Validity of the findings is the data’s ability to be
items. generalized across persons, settings, and times.
◦ Inter-item consistency reliability: the consistency of
respondent’s answers to all the items in a measure  Internal Validity related to the ability of a research
(Cronbach’s Alpha) instrument to measure what it is purported to measure.
◦ The split-half technique can be used when the measuring  Three major forms of internal validity:
tool has many similar questions.
Content validity
The instrument is administered and the results
separated into randomly selected halves. Criterion-related validity

If the correlation between the two halves are high, the Construct validity
instrument is highly reliable.

Bui Huy Hai Bich 03/11/2022 33 Bui Huy Hai Bich 03/11/2022 34
Research Methods for Business

Content Validity Criterion-Related Validity

 Is the extent to which the measurement instrument (predictive & concurrent validity)
provides adequate coverage of the investigative  Reflects the success of scales for prediction/estimation.
questions guiding the study.
 You may want to predict an outcome or estimate the
 To evaluate the content validity, one must first agree on existence of a current behavior.
what elements constitute adequate coverage.
◦ A scale that correctly forecasts the outcome of a purchase
 Assessment of content validity involves judgment. decision has predictive validity.

 First, by the designer , then by a panel. ◦ An observational method that correctly categorizes
families by current income class has concurrent validity.

Bui Huy Hai Bich 03/11/2022 35 Bui Huy Hai Bich 03/11/2022 36
Research Methods for Business

Construct Validity
 To evaluate construct validity, consider both
convergent and discriminant validity

 Convergent validity: the degree to which scores on

one scale correlate with scores on other scales
designed to assess the same construct.

 Discriminant validity: the degree to which scores on a

scale do not correlate with scores from scales
designed to measure different constructs.

Bui Huy Hai Bich 03/11/2022 37 Bui Huy Hai Bich 03/11/2022 38
Research Methods for Business

Practicality

- includes economy, convenience, and interpretability.

Economy:
 Data are not free; instrument length  costs. Contents…
 More items give more reliability  costs.
1. Nature of Measurement
 Data collection method costs.
2. Four basic measurement scales
Convenience:
 Easy to administer and clear in instructions.
3. Measurement of attitude
Interpretability: 4. Measurement errors
 When the designers ≠ interpreters of the results. 5. Good measurement

Bui Huy Hai Bich 03/11/2022 39 Bui Huy Hai Bich 03/11/2022 40