Marketing Research: Measurement & Scaling

• The chapter discusses concepts related to the measurement process, questionnaire design, sampling, and survey research.

Measurement Process

• Involves assigning numbers or labels to persons, objects, or events based on established rules.
• A rule is a guide, method, or command for researchers.

Steps in the Measurement Process

• Identify the concept of interest, which can be concrete or abstract.
• Develop a construct to simplify and integrate complex marketing phenomena.
• Define the concept constitutively by stating its meaning and distinguishing it in the study.
• Define the concept operationally by specifying measurable characteristics and assigning values.
• Develop a measurement scale using symbols or numbers assigned by rule.
• Measurement scale creation begins with determining the desired level.
• There are four basic levels of measurement: nominal, ordinal, interval, and ratio.

Levels of Measurement

• Nominal scales partition data into mutually exclusive and collectively exhaustive categories, involving classification-type data.
• Ordinal scales maintain labeling characteristics and allow ordering of data, involving ranking-type data.
• Interval scales have ordinal characteristics plus equal intervals, showing relative amounts, and allow comparison-type data.
• Ratio scales have interval characteristics plus a meaningful zero point, allowing flat numerical type data.

Evaluating Measurement

• Evaluation includes assessing the reliability and validity of the measurement.
• Accurate data implies accurate measurement, M = A (M = A + Error).
• Errors can be random or systematic.
• Systematic errors result in bias,while random errors influence measurement unsystematically.
• Scores on a measurement scale can differ due to characteristics being measured, stable characteristics, short-term factors, situational factors, survey variations or sampling of items.

Factors that cause differences in scores:

• Lack of clarity
• Instrument factors.
• Reliability is the degree to which measures are free from random error and provide consistent data.
• Validity is the degree to which an instrument measures the intended construct.

Role of a Questionnaire

• A questionnaire is a set of questions designed to generate necessary data

Criteria for a good questionnaire:

• Providing necessary decision-making information while considering the respondent.
• Considering the interviewing environment and length.
• Wording of the questionnaire is key to avoid bias.
• It should address editing, coding, and data analysis requirements.
• Skip patterns in questions should be based on answers.

Questionnaire Design Process Step 1

• Determine survey objectives, resources, and constraints.
• Objectives should be clear and precise.
• Resources include budget, time, and personnel, while constraints involve budget and time.

Questionnaire Design Process Step 2

• Determine the data collection method, such as in-person, self-administered, or internet survey.

Questionnaire Design Process Step 3

• Determine the question response format.
• Open-ended questions allow respondents to reply in their own words.
• Close-ended questions offer a list of answers.

Types of closed-ended questions:

• Dichotomous (2 options)
• Multiple-choice
• Scaled response responses.
• Coding is the process of grouping and assigning numeric codes to responses.
• Closed-ended questions are pre-coded.
• Open-ended questions are tedious and time-consuming to code.

Questionnaire Design Process Step 4

• Decide on the question wording, which must be clear and unbiased.
• Respondents must be able and willing to answer.

Consider factors such as:

• Respondents' knowledge
• Questionnaire length
• Sensitivity of questions

Questionnaire Design Process Step 5

• Establish questionnaire flow and layout.
• Use screening questions.
• Begin with general questions.
• Put sensitive, threatening, and demographic questions at the end.
• Use capital letters for instructions.
• Format questions appropriately, including screeners, warm-ups, transitions, complicated questions and demographics.

Questionnaire Design Process Step 6

• Evaluate the questionnaire.
• Ensure all questions are necessary.
• Consider the questionnaire's length and whether it generates required answers.

Questionnaire Design Process Step 7

• Obtain approval from relevant parties and distribute copies to those with authority over the project.

Final steps include:

• Make revisions
• Pretest and revise
• Prepare the final questionnaire
• Implement the survey.

Concept of Sampling

• It is a process of obtaining information from a subset of a larger group.
• The results are projected to the larger group.
• It is quicker and cheaper than a census
• Sampling a small percentage of a population can yield accurate population estimates
• Samples must be selected in a scientific manner.

Sample Versus Census

• Population is the entire group about whom information is needed.
• Census involves obtaining data from every member of the population and is rarely employed in marketing research.

Sample

• It is a subset of all members in a population of interest.
• Sampling makes estimates quicker and cheaper and enables accurate predictions for a large population based on a small sample.

Steps to Developing a Sampling Plan

• Define the population of interest
• Choose a data collection method
• Identify a sampling frame
• Select a sampling method (probability or non-probability)
• Determine sample size
• Develop operational procedures for selecting sample elements
• Execute the operational sample plan
• Defining the population of interest involves specifying characteristics of individuals or things needed to meet research objectives.
• Screener questions define the characteristics of individuals to be excluded.
• The researcher identifies the group to examine, study, or obtain information from.
• The main data collection methods include: personal, telephone, landline, mail, and internet.

Data Collection methods

• Mail suffers from low response rates.
• Telephone suffers from hang-ups.
• Internet suffers from a lack of adequate lists.

Sampling Frame

• A list is created of the members or elements of the population.
• It specifies a procedure to produce a representative sample.
• Sampling method selection depends on study objectives, resources, time, limitations, and the problem's nature.

Probability Sample

• Every element has a known, non-zero likelihood of selection.

Advantages of Probability Sample

• Information is obtained from a representative sample, sampling error can be computed, and results can be projected to the total population.

Disadvantages of Probability Sample

• It is more expensive to implement.
• Probabilities allow calculations of the extent to which a sample value can differ from a population value.

Non-probability Samples

• Elements are selected in a non-random manner, based on convenience.
• Used when sampling plant systematically excludes or over-represents a certain subset of the population.

Simple Random Sampling

• Selection is made by assigning a number to every element and using a table of random numbers to select elements.
• Formula = sample size / population size.

Systematic Sampling

• Used as a substitute for simple random sampling.
• The entire population is numbered, and elements are selected using a skip interval (population size / sample size).
• Is simpler, less timely and expensive compared to simple random sampling.

Stratified Sampling

• It is more representative through mutually exclusive and exhaustive subsets.
• The original population is divided into two or more mutually exclusive subsets.
• Random samples of elements from two or more subsets are chosen independently.
• The population is divided based on characteristics, making it statistically efficient with smaller sampling error.
• Stratified samples are used, despite statistical efficiency or not, due to the availability of required information and time consumption.

Allocation for data collection in stratified samples Includes

• Proportional: the number of elements selected from stratum is proportional to the stratum’s size relative to the population.
• Disproportional/Optimal: the number of elements taken from a given stratum is proportional to the stratum’s relative size and the standard deviation of the characteristics under consideration.
• Three steps to properly implement a proper stratified sample.
• Identify salient demographic or classification factors and determine proportions for subgroups under each stratum.

Cluster Sampling

• Sampling units are selected from small geographic areas to reduce costs.
• The population is divided into mutually exclusive subsets, and a random sample of subsets is selected.
• One-stage cluster samples consist of all elements in selected subsets.
• Two-stage cluster samples consist of elements chosen from selected subsets probabilistically.
• A sample frame is developed to specify groups or clusters, and the most popular type is geographic.
• It is statistically less efficient than random samples; certain sizes have a larger sampling error.

Area Sampling

• Geographic areas are selected that are progressively smaller.
• It encompasses steps using subdivisions, determining the geographic area, deciding on one or two step, using random selection, and employing a probability method.

Non-probability Sampling Methods

• Do not meet the requirements of a probability sample
• The main disadvantage is the inability to calculate sampling error.
• Four types of nonprobability samples: convenience, judgment, quota.
• Convenience samples are based on easily accessible people
• They are primarily used for convenience and are growing faster in exploratory situations.

Judgement Sampling

• Involves the researches selector

Quota Sampling

• Selected due to their demographic characteristics, or randomly, and classification factors are not selected based on correlation.

Snowball Sampling

• Procedure used to select additional respondents on the basis of referrals from initial respondents.
• Samples come from low-incidence or rare populations; used when finding respondents is extremely difficult; has the main advantage of being cost-efficient.

Internet Sampling

• Pros: respondents respond at their own convenience; it's inexpensive, survey software facilitates, and completion is quick.
• Cons: sample might not be representative, there may be uncertainty of the respondents or their anonymity, and has issues with data security.

Sampling Methods

• Considers probability.
• It is representative of the group, sample, and the results are projectable.
• It is expensive, and involves timely respondent selection.
• Considers non-probability
• The information is cheap and indicative

Data Collection Considerations

• Determine sample size.
• A sample size is a subset chosen to represent the population of interest.

Sample size depends on

• Homogeneity
• Response rate
• Incidence rate
• Number of subgroups.

Non-Probability Samples

• Rely on factors related to budget, time constraints, rules of thumb, and a number of subgroups.
• Budget availability determines affordability.
• Rules of thumb dictate convention and adequate sample size
• Client specifications.
• The best estimate is the "expert's".
• Draw on your experience and rely on convention.

Probability Samples

• Formulas are used to calculate sample size.
• Smaller sampling error = larger sample

Terms

• Acceptable error
• Level of confidence

Accuracy Factors

• Cost
• Managerial issues
• Research objectives

Statistical Methods

• Estimate standard deviation.
• Acceptable level of precision.
• Confidence intervals

Sampling Error

• The difference between sample value and the true value.
• Sampling error results when the selected sample is not perfectly representative, and it is bell-shaped, has one mode, and is symmetric.

Normal distribution factors

• Symmetric about mean
• Defined by mean
• Total area = 1

Normal Distribution Purpose

• Drawing inferences by making inferences on a single sample.
• Sample mean is best point estimate of the population mean.
• Intervals = range of values which true pop value does not fall.

Mean Sample Sizes

• Formula for calculating required sample size
• Formula factors include sampling error (E), level of confidence (Z), and population standard deviation (σ).

Sampling Distribution of the Proportion

• Relative frequency distribution of sample proportions from random samples; the mean proportion for possible samples equals the population proportion.

In Estimation Data

• There is no estimation data on sample or parameter

Population Parameters

• No direction between the population size and the estimated parameters
• It is only a parameter when it is large in the sample
• Rule of thumb: adjustment should be made in the sample size if the sample size is more than 5% of the size of the population.
• Selecting elements involves operational procedures; whether the sample is and specifying whether the sample used is probabilistic or not.
• Requirements include clear understanding
• Follow set process.
• Follow the specified procedures.

Why Use Surveys?

• To find out the "why".

Survey Research

• Structured approach to produce insights

Survey methods include:

• Interviewer administered
• Self-administered

Interview administration types:

Face to face

Telephone

Types of Self-administration

Online

Hand delivery

Post

Mobile

Intercept surveys

• Happens with consumers face to face

Mail surveys:

• Are cheap

Interview Surveys:

• Cons: The sample does not represent the population

Face to face surveys:

• Pros: Motivate
• Cons: Cost and time

Telephone Surveys:

Pros: Provide decentralized control Cons: Are limited by high rates and length

Online surveys

• Filled out without guidance from anyone
• They are limited in help and are easy to conduct with one less bias

Survey research

• Post (snail mail) has precontacted individuals who explain questionnaire.

Types of surveys:

• Cross-sectional
• Longitudinal

Telephone:

• Uses quantitative research

Uses of Online resources.

• Convenient
• incentivizing
• Providing 3 Elements to take into consideration when deciding which method to use. Budget, time, and accuracy Random error is the true value of error that is calculated.

Systematic Error factors

• Design
• Measurement
• Instruments
• Response rate
• Information
• Sample