Professional Documents
Culture Documents
Business Research Methods: Session 6 To 7
Business Research Methods: Session 6 To 7
Session 6 to 7
Prof. Vibhas Amawate
ERRORS IN RESEARCH
Errors in
Research
Design
Random
Systematic
Sampling
Error
Error
Administrative
Respondent
RANDOM SAMPLING ERROR
▪ There are many sources of systematic errors, which can be classified under
two broad categories, that is, administrative errors and respondent errors.
A D M I N I S T R AT I V E E R R O R S
▪ An error caused by improper administration or execution of the research task is known
as administration error. These are caused due to sample selection error or due to
other factors on the interviewer front such as carelessness, confusion, negligence,
omission, etc.
▪ Interviewer Error.
Error Details
Sample A systematic error that occurs because of an inaccuracy in either the
Selection Error stage of formulation of sample design or the execution of the
sampling procedure resulting in an unrepresentative sample is known as
sample selection error. E.g. Mall Intercept Interviewers selecting to
interview customers who are neatly dressed
Sample Frame The list of population elements or members from which units to be
Error sampled are selected is known as the sampling frame. A sampling frame
error is said to occur when this list of members does not correspond
exactly with the target population. E.g. List of all supermarkets in Delhi
excludes few
A D M I N I S T R AT I V E E R R O R S
Error Details
Population An error that results from an incorrect definition of the universe or
Specification population from which the sample chosen is known as a population
Error specification error. E.g. Looking of electric car potential in small car
buyers
Data Processing An error that occurs because of incorrect data entry, incorrect
Error computer programming or any other error during data analysis is
called data processing error.
Interviewer Error Interviewer error is an administrative error caused by mistakes
committed by the interviewer while administering the questionnaire
or recording the responses. E.g. Interviewer influencing the
respondent, Interviewer cheating etc.
A D M I N I S T R AT I V E E R R O R S
Error Details
Measurement • Measurement instrument error or bias occurs with the questionnaire
▪ Respondent error as the name suggests are those errors that are observed on the
respondents’ side. A survey requires the respondents’ cooperation in giving answers that
contain the correct information. In practice, it is very difficult to get the interviewees to
cooperate with the interviewer or reveal their true opinions.
▪ Non-Response : It is very difficult for any survey to achieve a 100 % response rate. The
statistical difference in results between a survey that includes only those who responded and
a perfect survey that would also include those who failed to respond is known as non-
response error.
▪ Response bias : A survey error that results from the inclination of people to answer a
question falsely, either through deliberate misrepresentation or unconscious falsification is
known as response bias.
RESPONDENT ERROR
▪ Deliberate Falsification : People might tend to give false answers in order to appear
intelligent or to conceal information they consider personal or embarrassing. Time pressure,
social desirability bias, courtesy bias and uninformed response errors are among other. e.g. a
respondent might remember the number of times he visited a supermarket in the last 6
months, but he might not be able to exactly recollect which supermarkets he visited and how
many times to each of them
▪ Unconscious Misrepresentation : Such situations can arise due to question format, content,
etc. It might happen that respondents misunderstand a question and give a wrong or biased
answer in the process. Prior inexperience to a subject or activity is also a reason why
unconscious misrepresentation on the part of the respondent occurs. Some respondents may
also consider it to be a prestige issue and try to answer every question thrown at them in the
best possible way rather than admitting that they do not know the answer to a question.
Overview of Measurement Scaling
▪ Once the research problem has been clearly established, the most important
part of the research, namely data collection, begins.
▪ First, it is not the person, object, or event that is being measured. It is an attribute or a
characteristic, such as the income of an individual, sales produced by an advertising campaign,
or degree of willingness of a person to purchase a product.
▪ Second, the numbers are assigned by rules and/or process specified by the researcher. While
the way the numbers are assigned can impact the types of analyses that can be done, the
process of assigning values is determined by the researcher. For instance, the number 1 does
not have to be assigned to females and 2 to males . So long as all females are designated by the
same number, and that number differs from the one assigned to males, the actual numbers used
are irrelevant
Types of Measurement Scales used in Research
▪ Researchers have four different types of scales that can be used for
measurements:
▪ Nominal
▪ Ordinal
▪ Interval
▪ Ratio
Nominal Scale
▪ With nominal scales numbers are assigned to objects or sets of objects for the purpose of
identification and classification. It is the lowest order of the four scales since the numbers
have no meaning other than identifying a particular characteristic. The number used for
each object or set is entirely up to the discretion of the researcher. The number have no
meanings.
▪ In developing nominal scales, it is highly desirable for the scale choices to be mutually
exclusive and collectively exhaustive.
▪ Mutually exclusive means that each response is uniquely different from other possible
responses. Thus, respondents fit into either one category or another, but cannot belong to
multiple categories.
▪ Categorically exhaustive means that all possible responses are included in the
answer categories.
Examples of Nominal Scales
▪ Gender: (1) Female (2) Male
▪ Preferred Fast Food: (1) McDonald's (2) Burger King (3) Wendy's (4) Taco Bell (5) KFC
▪ How did you hear about our business? (Check all that apply):
▪ _____ Radio
▪ _____ TV
▪ _____ Newspaper
▪ _____ Internet
▪ With ordinal scales numbers are assigned for the purpose of identification, but also have the
property of being arranged in some type of array or order. As with nominal scales, response
options should be mutually exclusive and categorically exhaustive. The sequence of
numbers means something in terms of one response category being larger or better than
another.
▪ Age: (1) 18–24 (2) 25–34 (3) 35–44 (4) 45–54 (5) 55+
▪ Number of websites visited: (0) None (1) 1 or 2 (2) 3 to 5 (3) 6 to 9 (4) 10 or more
▪ Please rank the following search engines in terms of your personal preference with 1 indicating your first choice, 2
indicating your second choice, 3 indicating your third choice, and 4 indicating your fourth choice.
▪ Bing: _____
▪ Google: _____
▪ WebCrawler: _____
▪ Yahoo: _____
Ordinal Scale Sample Reports
Interval Scales
▪ With interval scales the distance, or space between the numbers assigned to
objects or sets of objects, is considered to be equal.
▪ Interval scales subsume the properties of both the nominal and ordinal scales
in that numbers are assigned to objects or sets of objects and the numbers
chosen represent an ordered progression.
Interval Scales
▪ The difference between the ordinal and interval scales is that the distance
between scale points cannot be assumed to be equal with ordinal scales, while
with interval scales this assumption does hold true. While an ordinal ranking
question can tell us that something is ranked more favorably than something
else, we don't know by how much. The advantage of interval scales is that
they allow us to determine the degree to which something is preferred. For
this reason, interval scales are the scale of choice when measuring attitudes.
▪ It has characteristics of nominal and ordinal and also includes the power of
equality of interval
Interval Scales
▪ The difference between the ordinal and interval scales is that the distance between scale
points cannot be assumed to be equal with ordinal scales, while with interval scales this
assumption does hold true. While an ordinal ranking question can tell us that something is
ranked more favorably than something else, we don't know by how much. The advantage
of interval scales is that they allow us to determine the degree to which something is
preferred. For this reason, interval scales are the scale of choice when measuring attitudes.
▪ It has characteristics of nominal and ordinal and also includes the power of equality of
interval
Examples of Interval Scales
Please evaluate each of the following appetizers served at the Riverfront Grill
on a scale of 1 to 5 according to how well you like the appetizer.
Dislike a lot Somewh Neither Somewh Like a Lot
at Dislike Like nor at Like
Dislike
Cheese sticks 1 2 3 4 5
Coconut shrimp 1 2 3 4 5
Chicken fingers 1 2 3 4 5
▪ Because interval scales do not have an absolute zero point, in which 0 (zero) indicates a
total absence of the property being measured, you cannot compare means and say that one
is twice as much as another.
▪ For instance, if the mean for one appetizer is 4.0 and the mean for another one is 2.0, you
can't say that one appetizer is liked twice as much as the other. While the distances on an
interval scale are assumed to be equal, the absence of an absolute zero prevents such
comparisons.
Statistical Analysis on Different Scales
1. Nominal and Ordinal Scales are called non-metric scales and the data distribution would
be most likely be binomial and non-normal. The type of data analysis you could do is
limited
2. Interval and Ratio Scales are called metric scales and the data distribution would be
normal. All type of data analysis could be done.
Which Scale to Use When
▪ Relatively Easy for respondents to Understand : Scales should be relatively easy for
respondents to understand. Wording is important. It is advisable to use language that is
used by respondents, but also important to ensure the scale items are clear and concise.
Clarity and language familiarity are important, because these factors help to ensure that
respondents understand the question, and interpret it correctly, thus minimizing measurement
error.
▪ Clear and Concise : The scale needs to provide useful data, so in addition to being clear
and concise.
▪ Discriminates well between its Items : If a 5-point scale is used to measure attitude toward
a brand, and if 95% of the respondents check the same response category, then the question
does not discriminate adequately because it doesn't identify differences in attitudes.
Characteristics of a Good Scale
▪ Sample Selection Error: Scale items can be developed or borrowed from existing
scales that produce a wide range of responses. In rare cases when a given scale
does not discriminate well among respondents, the problem may be not with the
scale but with the sample selected. If avid Apple users are surveyed, they are very
likely to rate the Apple brand as a 5 on a 5-point rating scale.
Characteristics of a Good Scale
▪ Scales can be divided into two primary categories: comparative scales and non-comparative
Dichotomous A scale used to elicit a yes or Nominal E.g. do you own a TV-Yes or No
no answer.
Category Uses multiple item to elicit a Nominal Where in India do you reside?-
single response. Delhi, Mumbai, Chennai,
Bangalore, Others
Likert scale It measures how strongly Interval How strongly you agree or
subjects agree or disagree disagree with each of these
with statements on a 5 point statements on a five point
scale scale?
▪ Likert scales list a series of statements, and respondents are asked to indicate their level of
▪ One reason Likert scales are used frequently is their ease of construction.
▪ Likert scales avoid the difficulty of finding the right anchor words or phrases that are
▪ Likert scales are easy for respondents to read and understand. They are also easy to answer
as subjects can respond to a whole series of statements that utilize the same scale.
Likert Scales-Example
Likert Scales-Example
Likert Scales
▪ Likert scales can be either comparative or noncomparative. A comparative Likert scale may ask
individuals to respond to a series of statements about a particular retail store, such as the Gap, as it
compares to other clothing retail stores or a specific store.
▪ Likert scales often include both reversed (negatively phrased) and nonreversed (positively phrased)
items. Doing so is thought to reduce response bias due to acquiescent respondents who are likely to
be agreeable and rate everything positively
▪ Mixing positive and negative items is also important because some argue it helps to alert lazy or less
attentive respondents that the question content varies
▪ Unfortunately, reversing scale items has been found to diminish the reliability of scales and may result
in respondents selecting the exact opposite of their true attitude.
Main Categories of scales
Scale Definition Classification Examples
Numerical scale Is similar to Semantic Interval How pleased are you with the hostel
differential scale with a facilities?
difference that numbers Extremely Extremely
on a 5 or 7 point scales Pleased Displeased
are provided.
7 6 5 4 3 2 1
42
Semantic Differential Scales
▪ A semantic differential scale can be either comparative or non-comparative depending on
how the question is worded. The semantic differential scale involves a finite number of
choices anchored by dichotomous words or phrases. Most semantic differential scales have
5 or 7 points, which allows for a neutral position. In some cases the neutral position is
eliminated, and a 4- or 6-point scale is used to force respondents to choose one side or the
other of the scale. The key to good semantic differential scales is choosing the anchor
phrases or words that will produce discriminate answers among respondents. Scale anchors
should be bipolar, meaning that the anchors are perceived as “opposites” by
respondents.
▪ Semantic differentials are often used to develop brand and image profiles
Semantic Differential Scales-Example
Main Categories scales
Scale Definition Classification Example
Staple scale It measures both the direction and Interval scale Q. State how you would rate Maruti with
the intensity of the attitude towards respect to product innovation.
the item under the study. The -3 -2 -1 Maruti +1 +2 +3
characteristic of the study is placed
in the centre and a numerical scale
of +3 to -3 is indicated on both sides.
49
Stapel Scales
anchor can be expressed in positive terms, in neutral terms, or in a negative format. Whichever format
is used will impact the results. Using the anchor “value” produces different results than using anchors
▪ An additional disadvantage of the Stapel scale is that some respondents find it difficult to understand,
and may circle the anchor itself, rather than the positive or negative number that expresses the degree
to which the anchor reflects their positive or negative opinion. Misunderstandings of this nature are
more likely to occur when non-neutral anchors are used. Finally, Stapel scales may take up more
space on self-administered surveys, and thus be less desirable than semantic differentials. In practice,
Nor Unlikely 3
Somewhat Unlikely 2
Very Unlikely 1
55
Itemized Rating Scales
▪ With an itemized rating scale, respondents choose a response from a select number
of items or categories. It is also a noncomparative rating scale.
▪ The primary advantages of the itemized rating scale are it is easy for respondents to
understand and easy for researchers to code. No subjective judgment is needed to
identify what number should be given to a person's response. As a result, itemized rating
scales are used extensively in marketing research. They can be used in all types of survey
methods from telephone surveys to online surveys.
▪ Most itemized rating scales produce interval data because it is assumed there is equal
distance between each of the categories or items in the response.
Itemized Rating Scales-Example
Itemized Rating Scales-Example
Main Categories of attitudinal scales
Scale Definition Classification Example
Fixed or constant Here the respondents distribute Ordinal scale Q. In choosing a car indicate the importance
sum scale some points amongst the different that you attach to each of the following by
items. allotting points out of 100 in all.
59
Constant Sum
▪ The constant sum scale asks respondents to allocate points among various attributes
or brands to indicate their importance or preference relative to one another. Typically,
researchers will ask respondents to divide 100 points, but any number of points can be
used, such as 10 or 20. Because the total points must add to 100 (or another specified
number), the number of items to be ranked must remain small, usually under 10.
▪ The primary advantage of the constant sum scale over rank-order and the other comparative
scales is that the relative distance between rankings can be determined .
▪ Another advantage of constant sum scales is that they produce ratio data, which are the
highest order of data. In addition to rankings, researchers can compare the relative
magnitude of the rankings.
Constant Sum-Example
Constant Sum-Example
Main Categories of attitudinal scales
Scale Definition Classification Examples
Graphic rating A graphical representation helps respondents Ordinal On a scale of 1 to 10 how
scale to indicate their answer to a particular would rate the supervisor?
question. Excellent 10
All right 5
Very bad 1
63
Graphical Rating Scales
▪ Graphical rating scales are noncomparative scales that allow respondents to place
a response anywhere on a continuous line. Respondents are not making any
comparisons, but are asked to make a judgment or an evaluation. The scale is
normally anchored at each end with antonyms or words with highly different
meanings, such as poor and excellent or friendly and unfriendly.
▪ Because of the continuous nature of a graphical scale, it will produce interval data.
It cannot be ratio data since there is no absolute zero. Descriptive results would be
reported in terms of a mean and standard deviation.
Graphical Rating Scales- Example
COMPARITIVE SCALES
Comparative Scales
Scale Definition Classification
Paired comparison Among a small group of objects the respondents are asked to choose Ordinal
between two objects at a time. Thus if there are four cars you can
have =(n(n-1))/2=(4x(4-1))/2=6 paired comparisons.
Forced Ranking scale The respondent here ranks the different objects relative to each Ordinal
other. Lets say there are 5 brand to choose from. The respondent is
asked to assign a value of 1 to the least preferred choice and 5 to the
most preferred choice.
Comparative scale Often the researcher is interested in benchmarking provided the Interval scale
respondents are aware of the benchmarks. E.g How would you rate the
new Maruti Zen when compared with the earlier model?
1 3 5
67
Paired Comparisons
▪ Rather than ask individuals to rank order a set of items, researchers may use a series of
paired comparisons. With the paired comparison scale, respondents choose one of two
items in a set based on a specific criterion or attribute.
▪ The key advantage of the paired comparison scale is that it is typically easier for
respondents to choose between two items than it is to rank a series of items. The
paired comparison scale also tends to overcome order bias that may be created in listing
the items for a rank-order scale. Respondents may be influenced by the order in which
items are listed in the rank-order scale, even if the list is alphabetical.
▪ A major disadvantage is that its difficult to implement in case of too many comparisions.
Paired Comparisons- Example
Paired Comparisons- Example
Rank-Order Scales
▪ Respondent May not have knowledge or experience of all the items listed
▪ The research objectives comprise the research question/s or information areas and the
hypothesis
▪ Once the objectives and the research questions are identified, a researcher has to
develop a hypothesis statement that reflects these research objectives.
▪ If the objective of the research is to study the perceptions of the customer, a typical
research question could be: ‘Do the customers perceive the radiations from their
cell phones to be hazardous to health?’
Research Objectives, Questions and Hypothesis
Research Objectives Research Question Hypothesis
does not specify the variable to be investigated. Conversely, one that proposes that “Greater
▪ Research hypotheses that compare groups (users vs. nonusers, males vs. females, etc.) should
be stated in the plural form, and the subgroups to be compared specifically identified. “Women
rather than men are more likely to make a decision in purchase of refrigerators”
Features of a Correct Hypothesis
▪ The concept of null and alternative hypotheses is a basic statistics principle, although often
confusing.
▪ Null Hypothesis : Represented by the symbol H0, the null hypothesis is a statement or
claim that can be statistically tested. When the subject of a research hypothesis is a single
variable, the null hypothesis is stated in terms of equality. For instance, a research hypothesis
might propose that “Individuals most likely to purchase a yacht for personal use have an
average annual income of $5 million or more”.
▪ This research hypothesis would be rephrased into the following null hypothesis: “Individuals
most likely to purchase a yacht for personal use have an average annual income equal
to $5 million.”
Null and Alternate Hypothesis
that the questions are designed to draw information that will fulfil research
objectives.
▪ the target respondents : the researcher must make sure of the target population for
the survey. For example, in case of market research, a researcher has to decide
whether to include both users and non-users of a product or service. This is a
crucial step, as the sampling frame would be drawn after the target respondents
are defined.
Decisions to be considered before designing a questionnaire
▪ A researcher has to take many decisions before framing the actual
questionnaire. These decisions relate
▪ choice of survey or research method : The format and type of questions will be
different for personal interviews, focus groups, telephonic interviews and
mailed questionnaires
Factors to consider for Question Content
▪ Genuine ignorance about the topic : As respondents rarely admit the lack of knowledge of the
topic if you test there knowledge
▪ Inability to recollect the answer : What was your expenditure on grocery items in the last
week?’
▪ Inability to verbalize the response : This refers to the respondents’ inability to verbalize
factors influencing their buying motives. It is not quite possible to answer questions like, ‘Why
did you buy that car?’, ‘What made you buy that brand of shoes?’ This is because many
times people buy things for reasons other than what they admit to themselves.
Factors to consider for Question Content- The Respondent’s
Willingness to Answer Accurately
▪ This refusal can be because of the question being offending, too personal
and embarrassing, reflecting on prestige, or when the respondents
decide the topic is irrelevant to their interests.
Factors to consider for Question Content- External Content
▪ Example of external events is weather or time. For example, a questionnaire designed to find
the potential footfalls for a big apparel showroom that is to come up in a business district
investigated the shopping patterns of women by asking them ‘how many times did you go
shopping in the past 1 week?’. The survey was conducted just after a week of heavy rain in
the particular city. Though the answers were right for the particular question, it was not truly
representative of the shopping frequency of the respondents. Hence, questions should be
framed after considering external events.
Types of Questions –Open Ended Questions
▪ Q How would you evaluate Dhanraj Pillay’s performance in the Athens Olympics?
Types of Questions –Open Ended Questions
▪ Probing : Probing open-ended questions are those where the actual open-ended
questions are reached a little later in the process. Consider the following example
▪ Binary
▪ Ranking questions
▪ Multiple choice
▪ Checklist
Types of Questions –Close Ended Questions
▪ Binary Questions : These are also known as dichotomous questions as they permit only
two possible answers. The respondent has to choose one of the two permissible answers.
Binary Questions are mostly used in classification of the respondents.
▪ Ranking Questions : These questions require the participant to rank the response options
listed on a continuum basis in order of preference. Ranking questions are used to get
information that reveals participants’ attitudes and opinions. These questions list several
alternatives that might influence an individual’s decision-making.
Types of Questions –Close Ended Questions
▪ Three issues that should be considered while framing such questions are
▪ the position of the responses should be varied to avoid the selection of any
particular response due to position bias
▪ Checklist Questions : These are questions where the participant has the freedom to
choose one or more of the response options available. This is different from multiple-
choice questions in that it gives freedom to the respondents to choose one or more of
the options available.
Designing a Questionnaire-Question Wording
▪ Lead-in questions
▪ Qualifying questions
▪ Warm-up questions
▪ Specifics questions
▪ Demographics questions
Lead-in questions
▪ YES/NO
Qualifying Questions
▪ These are questions that slowly lead to the survey’s objective. This stage is
characterized by questions that evaluate the respondent and qualify him/her for
further questioning.
▪ A survey for estimating market potential for a new fluoride-based toothpaste brand
should ask qualifying questions like the following
▪ Questions like,
▪ ‘Looking back, can you recollect how many times you might have used fluoride toothpastes over
the last 1 year?’ tend to make the respondent think and recollect past experiences
Specific Questions
▪ This stage consists of questions that are specific to the research objectives.
▪ These specific questions play a major role in data collection and analysis.
After ensuring that enough rapport has been established, this section can
probe to gain insight into sensitive issues.
Demographic Questions
▪ These are a necessary part of every survey. Responses to survey questions cannot
be analysed until they are sorted out according to the different characteristics
pertaining to the study. This is especially true for surveys that analyse responses
based on the demographic characteristics of respondents. These usually consist of
a set of questions related to age, sex, location, occupation, etc. These questions
are kept to the end to avoid interviewee resistance and to prevent the interviewee’s
attention from being diverted.
Stages of Questionnaire Testing
▪ After the revision, the research instrument is ready for its final draft, which is
to be used for the actual survey.