π Lecture 13 β Sampling
π Overview: This lecture introduces the concept of sampling as a fundamental tool in marketing research. It explains when to use a sample versus a census, details the systematic sampling process, and covers the crucial steps of defining the target population and determining sample size. Understanding these concepts is essential for designing efficient and accurate marketing studies.
ποΈ Topics Covered
The lecture begins by defining population, census, and sample, and compares the conditions favoring the use of each. It then outlines the six-step sampling process, starting with reviewing research objectives and defining the target population in terms of elements, sampling unit, time, and extent. The process continues with identifying the sampling frame, determining sample size through various practical approaches, and selecting the sample.
π Lecture Summary
Sampling
Marketing research relies heavily on sampling, the process of selecting a portion of the population to represent the whole. Surveying the entire population (a census) is often impractical due to budget and time constraints.
π Definition β Population: All the elements which have the characteristics that we want to measure through the process of research; the entire group of study for the research at hand.
π Definition β Census: A complete counting of the population, allowing direct measurement of its parameters.
π Definition β Sample: A portion of the population that represents it; a representative subset selected for study. The sample unit is the basic level of investigation.
What should be used: Census or Sample
The choice depends on several factors. A census is feasible when: the population size is small; the research budget is plenty; the variance in the characteristic being measured is large; measurement is non-destructive; the study is not a secret; and the cost of sampling error is high. Conversely, a sample is appropriate when: the population size is large; available money is small; available time is short; sampling error cost is low; the measurement process is destructive; and variation in the characteristic is low. Sampling is more common in marketing research, offering benefits such as saving money and time, potentially greater accuracy (due to less non-sampling error), and being better when testing destroys the elements.
π‘ Why this matters: Understanding these conditions helps a researcher make a critical, practical decision that directly impacts the study's cost, speed, and validity.
Sampling Process
The sampling process consists of a series of sequential steps:
- Look at the research objective
- Define the population for study (elements, sampling unit, time, extent)
- Identify the sampling frame
- Determine sample size
- Select a sampling procedure
- Actually select the sample
1. Look at the Research Objectives
The research objectives, which include information needs, research questions, hypotheses, and boundaries, provide the foundation for every subsequent step, including defining the population and selecting the sampling technique.
2. Define the Population for Study
This is a basic and critical step. The target population must be specified precisely to identify the group from which the sample will be drawn. A population should be defined in terms of elements, sampling unit, time, and extent.
π Definition β Element: The object from which or about which the information is sought; the basis of analysis (e.g., individuals, families, companies).
π Definition β Sampling Unit: An element or a unit containing the element, selected at some stage of the sampling process (e.g., an individual, or a household containing that individual).
π Definition β Time: The time period under consideration (e.g., shoppers between October 01 to November 30, 2007).
π Definition β Extent: The geographical boundaries for the study (e.g., Lahore city).
π Example I: For a survey of female health care products, the target population is defined as:
- Element: Female 20-35
- Sampling unit: Female 20-35
- Time: Shopping between October 01 to November 30, 2007
- Extent: Lahore
π Example II: For a population to measure the reaction of a buyer to an industrial chemical:
- Element: Chemical Engineer
- Sampling unit: Chemical Engineers of companies purchasing over Rs. 5 million of chemicals per annum
- Time: 2006
- Extent: Punjab
π Example III: For a population to monitor the sale of a newly launched product:
- Element: Retail outlet where our product is sold
- Sampling unit: Departmental stores, general stores, medical stores, which sell our product
- Time: September 10 to 17, 2007
- Extent: Lahore city
Determine the Sampling Frame
A sampling frame is an up-to-date, clean, master list of all sampling units from which the sample will be drawn (e.g., a class list, telephone directory, or map). A sampling frame error occurs if this listing is inaccurate or outdated. Sometimes a perfect list is unavailable, so the researcher must edit or construct one, or define the frame by access (e.g., "all shoppers who buy over Rs. 500... during first week of October").
Determine the Sample Size
Sample size refers to the number of elements of the population to be included in the study. Several qualitative and practical approaches are used:
- Resource Constraints: Limited money, time, or personnel may push for a smaller sample. However, using budget to dictate sample size is considered "backward thinking"; the sample size should determine the budget.
- Comparable Studies: The researcher can check what sample sizes were used in similar studies that achieved a desirable level of reliability and then take an average.
- What the Client Says: The client may suggest a sample size based on their own quality or cost consciousness.
- Rule of Thumb: An arbitrary but practical guideline. For example, "it should be at least 3% of the population" or "at least 100 units in a major group and at least 30 in a subgroup."
β Key Takeaways
The fundamental distinction between a census (studying the entire population) and a sample (studying a representative subset) is crucial, with samples being far more common in marketing research due to practical constraints. The sampling process is a structured sequence starting with research objectives, which directly inform the precise definition of the target population using elements, sampling unit, time, and extent. An accurate sampling frame is essential to avoid frame error, and the sample size is determined by balancing resource constraints with practical heuristics like the "rule of thumb." The decision between sample and census depends on factors like population size, budget, and the destructive nature of measurement.
π§ Quick Revision Questions
- What are the four components used to define a target population for a research study?
- Under what conditions is it more appropriate to conduct a census rather than a sample?
- What is a sampling frame, and what error can arise if it is inaccurate or outdated?
- Name the six sequential steps in the sampling process as described in the lecture.
- What is the "rule of thumb" for sample size when the research involves major and subgroups?
π Lecture 14 β Other Factors Affecting the Sample Size
π Overview: This lecture discusses additional factors that influence sample size determination in marketing research beyond basic considerations. It introduces important statistical terms such as parameter, statistic, confidence level, precision level, and variability, and provides a step-by-step statistical method for calculating sample size with practical examples.
ποΈ Topics Covered
The lecture begins by examining other factors affecting sample size such as population variability and precision requirements. It then introduces key statistical terms including parameter, statistic, confidence level, precision level, variability, and corresponding Z values. The lecture provides methods for determining standard deviation, outlines steps for statistical sample size determination, presents the formula n = SΒ²ZΒ²/eΒ², and works through three detailed examples calculating sample size for different research scenarios.
π Lecture Summary
Other Factors Affecting the Sample Size
In addition to previous factors, population variability influences sample size. If the population is very heterogeneous, the sample must be divided into groups, requiring a larger sample size. Similarly, if research requires multivariate analysis, the sample size must be large. If the population is homogeneous and analysis is at the aggregate level, the sample can be small. Precision is another factor: if the decision based on research is important, data accuracy is a primary concern, calling for a larger sample to obtain data more precisely.
These concepts are used to determine sample size through the statistical method.
Some Important Statistical Terms
Parameter π Definition β Parameter: The summary measurement of the population or census.
Statistic π Definition β Statistic: The summary measurement of the sample, used to estimate the population parameter.
Confidence Level π Definition β Confidence Level: The probability with which we can say that our true value lies within a certain interval, called the confidence interval.
Precision Level π Definition β Precision Level: The allowable error between the parameter and the statistical value.
Variability π Definition β Variability: In statistics, variability is measured by variance or standard deviation. Standard deviation is the average distance of all units from the mean. In marketing research, usually 95% or 99% confidence level is used.
Corresponding 'Z' Values
- 'Z' value of 95% confidence level is 1.96
- 'Z' value of 99% confidence level is 2.58
- 'Z' value of 99.7% confidence level is 3.0
How to Determine Standard Deviation?
Standard deviation can be obtained from three sources:
- Taken from secondary data
- If you are an expert researcher, you can estimate the standard deviation
- Take a sample of 30 and measure its actual standard deviation
Steps for Determining the Sample Size Statistically
- Define precision level (e.g., Β±1, Β±2, etc.) β this varies from research to research according to objectives
- Define confidence level (e.g., 90%, 95%, etc.)
- Determine corresponding Z value for the confidence level
- Determine standard deviation
Symbols used:
- e = Precision level (error Β±)
- s = Standard deviation (Ο)
- z = Confidence level
- n = Sample size
π Formula: n = sΒ²zΒ² / eΒ²
Plain-English meaning: The required sample size equals the square of the standard deviation multiplied by the square of the Z value, divided by the square of the allowable error.
After using this formula, the answer may be in fraction form and should be rounded up.
Examples for Determining Sample Size
Example I π Example: We want to obtain the mean age of a magazine subscriber at 99% confidence level. We want our result to be Β±2 of the true mean age. Suppose standard deviation is 5. What should be the sample size?
Solution:
- e = Β±2
- s = 5
- z = 2.58
- n = sΒ²zΒ² / eΒ²
- n = (5)Β² Γ (2.58)Β² / (2)Β²
- n = 25 Γ 6.6564 / 4
- n = 41.6 or n = 42
π‘ Why this matters: Even with a relatively small standard deviation and modest precision, a 99% confidence level requires a larger sample size than lower confidence levels.
Example II π Example: A dairy company wants to determine the average consumption of milk per household that should be within Β±1.5 liters of the actual consumption. Past trend indicates that average variation in milk consumption per household is 4.5 liters. What should be the sample size if 95% confidence level is required?
Solution:
- e = Β±1.5
- s = 4.5
- z = 1.96
- n = sΒ²zΒ² / eΒ²
- n = (4.5)Β² Γ (1.96)Β² / (1.5)Β²
- n = 20.25 Γ 3.8416 / 2.25
- n = 34.5 or n = 35
Note: If we know the value of 'n' and are missing any one among precision level, standard deviation, or 'z' value, then the unknown value can also be determined by using the same formula: n = sΒ²zΒ² / eΒ²
Example III π Example: A researcher wants to find the monthly amount spent on lunch in restaurants. The researcher wants to be 95% confident that the results are within Β± Rs. 100. What will be the sample size if the standard deviation is estimated to be Rs. 400?
Solution:
- e = Β±100
- s = 400
- z = 1.96
- n = sΒ²zΒ² / eΒ²
- n = (400)Β² Γ (1.96)Β² / (100)Β²
- n = 160,000 Γ 3.8416 / 10,000
- n = 61.44 or n = 62
This shows that the researcher should take 62 people as sample in order to estimate the average income spent on lunch by professionals.
Key Relationships:
- When we increase the precision level, error decreases and sample size increases (and vice versa)
- When we increase the standard deviation, sample size increases
- When we increase the confidence level, sample size increases
β Key Takeaways
Sample size determination depends on population variability and required precision, especially for important decisions. The formula n = sΒ²zΒ²/eΒ² is the core statistical method, where 's' is standard deviation, 'z' is the Z value corresponding to the chosen confidence level (1.96 for 95%, 2.58 for 99%), and 'e' is the allowable precision error. Increasing precision, standard deviation, or confidence level all lead to larger required sample sizes. Standard deviation can be obtained from secondary data, expert estimation, or a pilot sample of 30. When calculating sample size, fractional results should be rounded up to the next whole number.
π§ Quick Revision Questions
- What are the three ways to determine standard deviation for sample size calculation?
- If a researcher wants to decrease the error from Β±2 to Β±1 while keeping everything else constant, what happens to the required sample size?
- Why is a sample size of 42 needed in Example I instead of just 41?
- What does it mean when we say "at 95% confidence level" in terms of the Z value and probability?
- If you know the sample size (n) but are missing the precision level (e), how can you find the missing value?
π Lecture 15 β Sampling Techniques
π Overview: This lecture introduces the two major categories of sampling techniques in marketing research: non-probability and probability sampling. It explains the definitions, procedures, advantages, and limitations of each technique, helping researchers understand when and how to select representative samples from a population.
ποΈ Topics Covered
The lecture covers the broad classification of sampling techniques into non-probability and probability sampling. Non-probability techniques discussed include convenience sampling, judgmental sampling, quota sampling, and snowball sampling. Probability techniques include simple random sampling, systematic sampling, and stratified sampling.
π Lecture Summary
Non-Probability Sampling Techniques
Non-probability sampling relies on the personal judgment of the researcher rather than chance to select sample elements. The researcher can arbitrarily or consciously decide what elements to include. While these methods may yield good estimates of population characteristics, they do not allow for objective evaluation of the precision of sample results, and the estimates obtained are not statistically projectable to the population.
Convenience Sampling
Convenience sampling attempts to obtain a sample of convenient elements. The selection of sampling units is left primarily to the interviewer. Examples include use of students, church groups, and members of social organizations; mall intercept interviews without qualifying respondents; department stores using charge account lists; tear-out questionnaires in magazines; and "people on the street" interviews.
Convenience sampling is the least expensive and least time-consuming of all sampling techniques. The sampling units are accessible, easy to measure, and cooperative. In spite of these advantages, this form of sampling has serious limitations. Many potential sources of selection bias are present, including respondent self-selection. Convenience samples are not representative of any definable population.
π Definition β Selection Bias: Bias that occurs when the procedure used to select the sample results in a sample that is not representative of the population.
Hence, it is not theoretically meaningful to generalize to any population from a convenience sample, and convenience samples are not appropriate for marketing research projects involving population inferences. Convenience samples are not recommended for descriptive or causal research, but they can be used in exploratory research for generating ideas, insights, or hypotheses. Convenience samples can be used for focus groups, pretesting questionnaires, or pilot studies. Nevertheless, this technique is sometimes used even in large surveys.
Judgmental Sampling
Judgmental sampling is a form of convenience sampling in which the population elements are selected based on the judgment of the researcher. The researcher, exercising judgment or expertise, chooses the elements to be included in the sample because he or she believes they are representative of the population of interest or are otherwise appropriate. Common examples include: test markets selected to determine the potential of a new product; purchase engineers selected in industrial marketing research; expert witnesses used in court; and department stores selected to test a new merchandising display system.
Judgmental sampling is low cost, convenient, and quick, yet it does not allow direct generalizations to a specific population. Judgmental sampling is subjective, and its value depends entirely on the researcher's judgment, expertise, and creativity. It may be useful if broad population inferences are not required.
Quota Sampling
Quota sampling may be viewed as two-stage restricted judgmental sampling. The first stage consists of developing control categories, or quotas, of population elements. The relevant control characteristics, which may include sex, age, and race, are identified on the basis of judgment. In other words, the quotas ensure that the composition of the sample is the same as the composition of the population with respect to the characteristics of interest. In the second stage, sample elements are selected based on convenience or judgment.
Snowball Sampling
In snowball sampling, an initial group of respondents is selected, usually at random. After being interviewed, these respondents are asked to identify others who belong to the target population of interest. Subsequent respondents are selected based on the referrals. This process may be carried out in waves by obtaining referrals from referrals, thus leading to a snowballing effect. A major objective of snowball sampling is to estimate characteristics that are rare in the population. Snowball sampling is used in industrial buyer-seller research to identify buyer-seller pairs. The major advantage of snowball sampling is that it substantially increases the likelihood of locating the desired characteristic in the population.
Probability Sampling Techniques
In probability sampling, sampling units are selected by chance. Every potential sample need not have the same probability of selection, but it is possible to specify the probability of selecting any particular sample of a given size.
Simple Random Sampling
In Simple Random Sampling (SRS), each element in the population has a known and equal probability of selection. Every element is selected independently of every other element. The sample is drawn by a random procedure from a sampling frame. This method is equivalent to a lottery system in which names are placed in a container, the container is shaken, and the names of the winners are then drawn out in an unbiased manner.
π Definition β Sampling Frame: A list of all the elements in the population from which the sample is drawn.
The researcher first compiles a sampling frame in which each element is assigned a unique identification number. Then random numbers are generated to determine which elements to include in the sample.
π Formula: Sampling interval (i) = Population size (N) / Sample size (n) β This determines how often to select elements in systematic sampling.
SRS has many desirable features. It is easily understood. The sample results may be projected to the target population. SRS suffers from at least four significant limitations:
- It is often difficult to construct a sampling frame.
- SRS can result in samples that are very large or spread over large geographic areas, increasing time and cost of data collection.
- SRS often results in lower precision with larger standard errors than other probability sampling techniques.
- SRS may or may not result in a representative sample. Although samples drawn will represent the population well on average, a given simple random sample may grossly misrepresent the target population.
For these reasons, SRS is not widely used in marketing research.
Systematic Sampling
In systematic sampling, the sample is chosen by selecting a random starting point and then picking every i-th element in succession from the sampling frame. The sampling interval, i, is determined by dividing the population size N by the sample size n and rounding to the nearest integer.
Systematic sampling is less costly and easier than SRS, because random selection is done only once. Moreover, the random numbers do not have to be matched with individual elements as in SRS. Systematic sampling is often employed in consumer mail, telephone, mall intercept, and internet interviews.
Stratified Sampling
Stratified Sampling is a two-step process in which the population is partitioned into sub-populations, or strata. The strata should be mutually exclusive and collectively exhaustive in that every population element should be assigned to one and only one stratum and no population elements should be omitted. Next, elements are selected from each stratum by a random procedure, usually SRS.
Stratified sampling differs from quota sampling in that the sample elements are selected probabilistically rather than based on convenience or judgment. A major objective of stratified sampling is to increase precision without increasing cost.
π Definition β Strata: Sub-populations within a larger population that are partitioned for sampling purposes.
The elements within a stratum should be as homogeneous as possible, but the elements in different strata should be as heterogeneous as possible. Variables commonly used for stratification include demographic characteristics, type of customer (credit card versus non-credit card), size of firm, or type of industry.
Stratified sampling can be proportionate or disproportionate.
π‘ Why this matters: Stratified sampling is one of the most efficient probability techniques. By ensuring that all important subgroups are represented, it can provide more precise estimates than simple random sampling for the same sample size.
β Key Takeaways
The most critical distinction in sampling is between non-probability and probability techniques. Non-probability sampling relies on researcher judgment, is faster and cheaper, but does not allow for statistical projection to the population. Probability sampling uses chance selection, permits objective evaluation of precision, and allows for population inference. Key techniques include convenience (easiest but least representative), judgmental (expert-based but subjective), quota (two-stage restricted judgmental), and snowball (for rare populations) for non-probability, and SRS (equal probability but difficult to implement), systematic (sampling interval), and stratified (partitioning into homogeneous subgroups for increased precision) for probability. Stratified sampling is often preferred for precision, while systematic sampling is more practical than SRS.
π§ Quick Revision Questions
- What is the fundamental difference between non-probability and probability sampling?
- What are the main limitations of convenience sampling, and when can it be appropriately used?
- How does quota sampling differ from stratified sampling?
- What is a sampling frame, and why is it essential for Simple Random Sampling?
- What is the major objective of stratified sampling, and what property should elements within a stratum exhibit?
π Lecture 16 β Stratified Sampling
π Overview: This lecture covers advanced probability sampling techniques, focusing on stratified and cluster sampling methods. It explains how these techniques differ in objectives, implementation, and when each is appropriate for marketing research, while also comparing various probability and non-probability sampling methods.
ποΈ Topics Covered
The lecture explains stratified sampling (both proportionate and disproportionate), cluster sampling and area sampling, other techniques like sequential and double sampling, a comparison table of all sampling methods, mall intercept sampling challenges, and sampling errors including sampling frame errors.
π Lecture Summary
Stratified Sampling
Stratified Sampling is a two-step process where the population is partitioned into sub-populations called strata. Strata must be mutually exclusive and collectively exhaustive, meaning every population element belongs to exactly one stratum and no elements are omitted. Elements are then selected from each stratum using a random procedure, usually Simple Random Sampling (SRS). This differs from quota sampling because elements are selected probabilistically, not by convenience or judgment. A major objective of stratified sampling is to increase precision without increasing cost.
π Definition β Strata: Sub-populations within a population that are mutually exclusive and collectively exhaustive, created to improve sampling precision.
Elements within a stratum should be as homogeneous as possible, while elements in different strata should be as heterogeneous as possible. Common stratification variables include demographic characteristics, type of customer (credit card vs. non-credit card), firm size, or industry type. Stratified sampling can be either proportionate or disproportionate.
Proportionate Stratified Sampling
In proportionate stratified sampling, the sample size from each stratum is proportional to that stratum's size in the population.
π Example: Population sizes: A=450, B=350, C=100, D=250, E=50 (Total=1200). Sample sizes: A=90, B=70, C=20, D=50, E=10 (Total=240). Each sample size is exactly 20% of its stratum population (e.g., 90/450 = 20%).
Disproportionate Sampling
In disproportionate sampling, the sample size is not proportional to the population size of each stratum.
π Example: Population sizes: A=450, B=350, C=100, D=250, E=50 (Total=1200). Sample sizes: A=70, B=60, C=30, D=50, E=30 (Total=240). Here, stratum C (population 100) gets 30 instead of 20, while stratum A (population 450) gets 70 instead of 90.
π‘ Why this matters: Disproportionate sampling is useful when certain strata have small populations but are critically important, or when some strata have greater variability and need larger samples.
Cluster Sampling
In cluster sampling, the target population is first divided into mutually exclusive and collectively exhaustive subpopulations called clusters. The key distinction from stratified sampling is that in cluster sampling, only a sample of clusters is chosen for further sampling, whereas in stratified sampling, all strata are selected.
The objectives differ: cluster sampling aims to increase sampling efficiency by decreasing costs, while stratified sampling aims to increase precision. Elements within a cluster should be as heterogeneous as possible, but clusters themselves should be as homogeneous as possible β the opposite of stratified sampling.
A common form is area sampling, where clusters consist of geographic areas like counties, housing tracts, or blocks. In one-stage area sampling, all households in selected blocks are included.
Advantages of cluster sampling: feasibility and low cost. Often, only cluster-level sampling frames are available, not element-level frames. It is the most cost-effective probability sampling technique.
Limitations: Cluster sampling produces relatively imprecise samples, and it is difficult to form heterogeneous clusters because households in a block tend to be similar.
Other Probability Sampling Techniques
Sequential Sampling: Population elements are sampled sequentially, with data collection and analysis at each stage. A decision is made whether to continue sampling. The sample size is not known in advance, but a decision rule is stated before sampling begins, indicating when enough information has been obtained.
Double Sampling (also called two-phase sampling): Certain population elements are sampled twice.
Comparison of Various Sampling Techniques
Probability Sampling Strengths and Weaknesses:
| Method | Strengths | Weaknesses |
|---|---|---|
| Simple Random Sampling | Projectable results, easy to understand | Hard to construct sampling frame, expensive, low precision, representativeness uncertain with high variability |
| Systematic Sampling | Can work without sampling frame, easier implementation | Representativeness can decrease |
| Stratified Sampling | More precision, handles all subpopulations' variability | Stratification variables hard to identify, difficult with many variables, more expensive |
| Cluster Sampling | Economical in cost and time | Lacks precision, difficult computation and interpretation |
Non-Probability Sampling Strengths and Weaknesses:
| Method | Strengths | Weaknesses |
|---|---|---|
| Convenience Sampling | Economical, less time-consuming, convenient | Researcher bias, lack of representativeness, not recommended for conclusive research |
| Judgmental Sampling | Low cost, convenient, least time-consuming | Subjective results, very tentative generalization |
| Quota Sampling | Sample controllable for certain characteristics | Selection bias, lack of representativeness |
| Snowball Sampling | Easy identification of relevant units | Very time-consuming |
Mall Intercept Sampling
Mall intercept sampling (shopping center sampling) presents unique difficulties. Bias may be introduced by three factors:
-
Selection of shopping center: A shopping center reflects the local area's family type (e.g., low vs. high income). Using several centers in different areas improves representativeness. If possible, select cities with diversity.
-
Selection of location within the shopping center: The part of the mall (beginning, center, or end) affects representativeness. All locations should be used in rotation. Different parking lot entrances matter. If the interviewing facility is far, refusals may increase due to time constraints.
-
Part of the day (time) selection: Different people shop at different times β women on weekdays, working people in evenings, families on weekends, working people at lunch hour. Evenings are usually peak hours. Samples should be taken at different hours.
Sampling Errors
Sampling error is the difference between the sample value (statistic) and the true underlying population value (parameter) . It occurs due to sample size and sampling technique. Some methods can minimize sampling error.
A sampling frame error is the degree to which the sampling frame (or list) fails to account for the population. For example, Yellow Pages and telephone directories often contain sufficient sampling frame errors because they are not up-to-date β some numbers are redundant while others are not listed.
π Definition β Sampling Error: The difference between the sample statistic and the true population parameter, caused by sample size and sampling technique. π Definition β Sampling Frame Error: The degree to which the sampling frame fails to accurately represent the target population.
β Key Takeaways
Stratified sampling increases precision by creating homogeneous strata and sampling all of them, while cluster sampling reduces costs by sampling only some clusters with heterogeneous elements within them. Proportionate stratified sampling matches population proportions, while disproportionate sampling allows different strata weights. Mall intercept sampling requires careful consideration of center selection, location within center, and time of day to minimize bias. Sampling error is the difference between sample and population values, while sampling frame error occurs when the list used does not accurately represent the population. Understanding the trade-off between cost, precision, and representativeness is critical for choosing the appropriate sampling technique.
π§ Quick Revision Questions
- What is the key difference in objectives between stratified sampling and cluster sampling?
- In stratified sampling, should elements within a stratum be homogeneous or heterogeneous?
- Give an example of a variable commonly used for stratification in marketing research.
- Why is mall intercept sampling prone to bias, and what three factors must be carefully considered?
- What is the difference between sampling error and sampling frame error?
π Lecture 17 β Development of Questionnaire
π Overview: This lecture focuses on the systematic process of developing questionnaires and observation forms for marketing research. It explains how to design questions that gather accurate, consistent data while motivating respondents to participate, covering everything from question content to overcoming respondent limitations.
ποΈ Topics Covered
The lecture covers the definition and objectives of questionnaires, the questionnaire design process starting with specifying information needed, determining the type of interviewing method, individual question content including necessity and double-barreled questions, overcoming inability to answer (informedness, memory, articulation), and overcoming unwillingness to answer (effort, context, legitimacy, sensitivity) with techniques to increase respondent willingness.
π Lecture Summary
Questionnaire and Observation Forms
Survey and observations are the two basic methods for obtaining quantitative primary data in descriptive research. Both require a procedure for standardizing the data collection process so that data are internally consistent and can be analyzed uniformly. If 40 different interviewers conduct personal interviews or make observations in different parts of the country, the data they collect will not be comparable unless they follow specific guidelines and ask questions and record answers in a standard way. A standardized questionnaire or form will ensure comparability of the data, increase speed and accuracy of recording, and facilitates data processing.
Questionnaire Definition
A questionnaire, whether called a schedule, interview form, or measuring instrument, is a formalized set of questions for obtaining information from respondents.
Objectives of a Questionnaire
Any questionnaire has three specific objectives:
- First, it must translate the information needed into a set of specific questions that the respondents can and will answer.
- Second, a questionnaire must uplift, motivate, and encourage the respondent to become involved in the interview, to cooperate, and to complete the interview. A well designed questionnaire can motivate the respondents and increase the response rate.
- Third, a questionnaire should minimize response error, defined as the error that arises when respondents give inaccurate answers or their answers are miss-recorded or miss-analyzed. A questionnaire can be a major source of response error.
Questionnaire Design Process
The lecture presents a flow diagram of the questionnaire design process with the following steps:
- Specify the Information Needed
- Specify the Type of Questioning Method
- Design the Questions
- Determine Structure of the Question
- Determine the Question Wording
- Identify Layout of the Questionnaire
- Determine the Sequence of Questions
- Produce the Questionnaire Form
- Do Pre-testing
Specify the Information Needed
The first step in questionnaire design is to specify the information needed, which is also the first step in the research design process. It is helpful to review components of the problem and the approach, particularly the research questions, hypotheses, and the information needed. It is also important to have a clear idea of the target population. The characteristics of the respondents group have a great influence on questionnaire design. Questions that are appropriate for college students may not be appropriate for housewives. Understanding is related to respondent socioeconomic characteristics. The more diversified the respondent group, the more difficult it is to design a single questionnaire that is appropriate for the entire group.
Type of Interviewing Method
An appreciation of how the type of interviewing method influences questionnaire design can be obtained by considering how the questionnaire is administered under each method. In personal interviews, respondents see the questionnaire and interact face to face with the interviewer, so lengthy, complex, and varied questions can be asked. In telephone interviews, the respondents interact with the interviewer but do not see the questionnaire, which limits questions to short and simple ones. Mail questionnaires are self-administered, so the questions must be simple and detailed instructions must be provided.
Individual Question Content
Is the Question Necessary?
Every question in a questionnaire should contribute to the information needed or serve some specific purpose. If there is no satisfactory use for the data resulting from a question, that question should be eliminated. It is useful to ask some neutral questions at the beginning of the questionnaire to establish involvement and rapport, particularly when the topic is sensitive or controversial. Sometimes filler questions are asked to disguise the purpose or sponsorship of the project. Rather than limiting the questions to the brand of interest, questions about competing brands may also be included to disguise the sponsorship.
Are Several Questions Needed Instead of One?
A double-barreled question combines two or more questions into one and is incorrect. For example: "Do you think Cola A is a tasty and refreshing soft drink?" is incorrect because it asks about taste and refreshment together. To obtain the required information, two distinct questions should be asked:
- "Do you think Cola A is a tasty soft drink?"
- "Do you think Cola A is a refreshing soft drink?" (Correct)
Overcoming Inability to Answer
Researchers should not assume that respondents can provide accurate or reasonable answers to all questions.
Is the Respondent Informed?
A husband may not be informed about monthly expenses for groceries and department store purchases if it is the wife who makes these purchases, or vice versa. Research has shown that respondents will often answer questions even though they are uninformed. In situations where not all respondents are likely to be informed, filter questions that measure familiarity, product use, and past experience should be asked before questions about the topic themselves. Filter questions enable the researcher to filter out the respondents who are not adequately informed.
Can the Respondent Remember?
Many things that we might expect everyone to know are remembered by only a few. Examples of questions that exceed memory ability include:
- "What is the brand name of the trouser you were wearing three weeks ago?"
- "What did you have for dinner two weeks ago?"
- "How many liters of soft drinks did you consume during the last three weeks?"
These questions are incorrect as they exceed the ability of respondents to remember. Evidence indicates that consumers are particularly poor at remembering quantities of products consumed. Soft drink consumption may be better obtained by asking frequency questions with categories:
- Less than once a week
- One to three times per week
- Four to six times per week
- Seven or more times per week
Can the Respondent Articulate?
Respondents may be unable to articulate certain types of responses, for example, the atmosphere of a department store. If respondents are unable to articulate their responses to a question, they are likely to ignore that question and may refuse to respond to the rest of the questionnaire. Respondents should be given aids such as pictures, maps, and descriptions to help them articulate their responses.
Overcoming Unwillingness to Answer
If respondents are able to answer a particular question, they may be unwilling to do so because too much effort is required, the context may not seem appropriate for disclosure, no legitimate purpose or need for the information is apparent, or the information requested is sensitive.
Effort Required of the Respondents
The researcher should minimize the effort required of the respondents. Suppose the researcher is interested in determining from which departments in a store the respondent purchased merchandise on the most recent shopping trip. Asking respondents to list all departments (open-ended) is incorrect. A better approach is to provide a checklist where respondents can check all departments, such as:
- Womenβs shoes
- Menβs apparel
- Childrenβs apparel
- Cosmetics
- Jewelry
- Other (please specify)
The second option is preferable because it requires less effort from respondents.
Context
Some questions may seem appropriate in some contexts but not in others. For example, questions about personal hygiene habits may be appropriate when asked in a survey sponsored by the Pakistan Medical Association, but not in one sponsored by a fast food restaurant. Respondents are unwilling to respond to questions they consider inappropriate for the given context.
Legitimate Purpose
Respondents are also unwilling to divulge information that they do not see as serving a legitimate purpose. For example: "Why should a firm marketing cereals want to know their age, income, and occupation?" A statement such as, "To determine how the consumption of cereal and preferences of cereal brands vary among people of different ages, incomes, and occupations, we need information on..." can make the request for information seem legitimate.
Sensitive Information
Respondents are unwilling to disclose, at least accurately, sensitive information because this may cause embarrassment or threaten the respondent's prestige or self-image. Sensitive topics include money, family life, political and religious beliefs, and involvement in accidents or crimes.
Increasing the Willingness of Respondents
Two techniques to increase willingness:
- Place sensitive topics at the end of the questionnaire. By then, initial mistrust has been overcome, rapport has been created, legitimacy of the project has been established, and respondents are more willing to give information.
- Preface the question with a counter biasing statement β a statement that the behavior of interest is common. For example, before requesting information on credit card debt, say, "Recent studies show that most Pakistanis are in debt."
β Key Takeaways
Questionnaire design is a systematic process that begins with specifying the information needed and understanding the target population. The three key objectives of any questionnaire are translating information needs into answerable questions, motivating respondent participation, and minimizing response error. Researchers must design questions that overcome both inability to answer (through filter questions, memory aids, and articulation support) and unwillingness to answer (by reducing effort, ensuring context-appropriateness, establishing legitimate purpose, and handling sensitive topics carefully). The interviewing method used (personal, telephone, or mail) directly influences question complexity and structure. Finally, counter biasing statements and placing sensitive questions at the end are proven techniques to increase respondent willingness to disclose difficult information.
π§ Quick Revision Questions
- What are the three specific objectives of any questionnaire?
- What is a double-barreled question, and why should it be avoided?
- What is a filter question, and when should it be used?
- What is a counter biasing statement, and how does it help overcome unwillingness to answer?
- How does the type of interviewing method (personal vs. telephone vs. mail) influence questionnaire design?