MARKETING RESEARCH EXAM 2
all of the above (which is: once, never, often)
Which of the words below is an ambiguous word?
Stapel scale.
Which scale typically has numbers that range from a minus end (-3) to a corresponding plus end (+3) and includes a 0 midpoint?
All of the above should be avoided. (Which is: Have you ever seen Lost? ; Do you always watch Lost? ; Have you never watched Lost? ; Have you ever watched and recorded an episode of Lost?)
Which of the following is the wrong way to ask if someone has viewed an episode of Lost?
A single word can make a difference in how respondents respond to a question.
Which of the following is true and should be remembered when developing questions?
All of the above. (which is: Questionnaire translates research objectives into specific questions. ; Questionnaire standardizes specific questions and response categories. ; Questionnaire serves as a permanent record of research. ; Through wording and question flow, it fosters respondent cooperation.)
Which of the following functions does a questionnaire serve?
Have you heard about the satellite radio system?
Which of the following is NOT a leading question?
Demographic questions
Which of the following is NOT included in the introduction part of the questionnaire?
They are easier and cheaper.
Which of the following is a distinct advantage of computer-assisted questionnaire design software packages?
Close-ended questions give respondents some degree of control over the answers.
Which of the following is not the benefit of using closed-end questions:
avoiding ambiguous words
When developing a questionnaire, if the researcher is asking him/herself if the word means what he/she intended; if it has any other meanings; if the word has more than one pronunciation; or if a simpler phrase is suggested, then he/she is probably concerned with ________.
One-way labeled scale.
When you want to measure the frequency of use, which of the following scales is recommended?
The funnel approach.
Which approach to organizing the flow of questions would order questions from general to specific or from wide to narrow?
categorical
Which level of measurement is represented by answers on a scale, such as "yes," "no" or "male," "female" represent which level of measurement?
Because it cannot be measured, researchers must know the sources of non-sampling error and learn how to minimize it.
Which of the following best represents what should be done about non-sampling error?
The size of the sample.
Which of the following causes sampling error?
all of the above. (which is: focused, grammatically simple, brief, crystal clear)
According to the rules of question wording, a question should be:
sample has less sampling error than a census
All of the following are reasons why a researcher might choose a sample over a census for making inferences about the population, except
Always.
In the question, "Do you always buy electronic products from Dell?" which word could cause bias in respondents' answers?
Most.
In the question, "What was the most important factor that convinced you it was time to make this purchase?" which word could cause bias in respondents' answers?
open-ended
"Describe your college experience" is ________ questions.
double-barreled question
"Do you think Coca-Cola is a tasty and refreshing soft drink?" is an example of a ________.
multiple-choice categorical question.
"How many times have you visited the library in the last month? A) once; B) twice; C) more than twice" is an example of:
loaded
"How much do you think you would pay for a pair of sunglasses that will protect your eyes from the sun's harmful ultraviolet rays, which are known to cause blindness?" is an example of a(n) ________ question.
loaded
"Should people be allowed to protect themselves from harm by using mace as self-defense?" is an example of a(n) ________ question.
All of the above. (which is: Behavior, State of Being, State of Mind)
What information of consumers can be collected using questionnaire?
census
A ________ is defined as an accounting of everyone in the population.
conditional branching or filtering
A ________ question is one whose answer affects which question will be asked next.
Ordinal
A bank asks customers to evaluate the drive-thru service as good, average, or poor. What data format is this classification?
poses two different questions in one question and should be avoided.
A double-barreled question is one that:
is worded in such a way that gives the respondent a clue as to how to answer and should be avoided.
A leading question is one that:
synthetic scale.
A question asks respondents to evaluate a travel agency on a scale from 1 to 5 where 1 means "very dissatisfied" and 5 means "very satisfied." This is an example of a:
sampling error.
A researcher investigating the consumption of sugar takes a random sample from the population and computes the mean sugar consumption. He repeats the procedure a number of times using the exact same procedure and finds the means obtained vary from sample to sample. This is an example of
Questionnaire.
A researcher may be interested in knowing what they think about the promotional display in a supermarket and how they behave when exposed to a display. In this case, the researcher would want to ask consumers a bunch of questions about it. What form would the researcher use?
be representative of the entire population.
A sample is a subset of a group that should:
artificial numbers to measure consumers' state of mind.
A synthetic metric question uses:
open-ended
A(n) __________ question allows respondents to express opinions, ideas, or behaviors in their own words without being forced to choose among alternatives that have been predetermined by a marketing researcher.
closed-ended
A(n) __________ question requires respondents to select one or more response options from a set of predetermined choices.
semantic differential
A(n) __________ scale is a five-point scale in which the opposite ends have one- or two-word adjectives that have opposite meanings.
Likert
A(n) __________ scale is one in which the respondent indicates the extent to which he or she agrees or disagrees with a statement.
What is personal fashion style?
Before opening six Torrid plus-size-only retail stores that cater to women aged 15-30, a great deal of information was gathered to determine what types of items should be carried, the image of the store, its advertising, etc. Which of the following is an example of an open-end question that might have been asked?
specify the information needed
Before the questionnaire design, we must first ________.
Measurement; Scaling
Consider a scale from 1 to 100 for locating consumers according to the characteristic "attitude toward department stores." Each respondent is assigned a number from 1 to 100 indicating the degree of (un)favorableness, with 1 = extremely unfavorable, and 100 = extremely favorable. ________ is the actual assignment of a number from 1 to 100 to each respondent. ________ is the process of creating a 100-point favorability scale for measuring their attitude toward department stores.
non-sampling errors.
Errors that pertain to anything in the research process except sample size are known as:
interval data
If the distance between the "2" and the "3" and the "6" and the "7" on a scale are equal, the data format would likely be at least:
categorical, metric
In a survey for ACME Electric Co., Ralph Thomas asks the following questions: "What is your gender? Male-Female" and "From 1 to 10, how much do are you satisfied with the service you received?" In the first question regarding gender, the answers are composed of a small number of distinct values or categories. In the second question, the answers are composed of numbers that have an "underlying measurement continuum." How would you classify these two questions?
Screens are first, warm-ups second, and classification questions are last.
In terms of locating types of questions on the questionnaire, which of the following is true?
must be identified and appropriate steps must be taken to limit or eliminate their occurrence.
Jane Ellen Roberts is concerned about error in a survey that she is conducting for her company. Specifically, she is concerned about non-sampling errors but she realizes non-sampling errors:
have an impact on the response rate in that only those respondents who pass the screening will be included in the final sample.
Margaret Day Craddock is the director of MIFA, a charitable organization in Memphis, Tennessee. Margaret has contracted with a marketing research firm in Memphis to conduct a telephone survey of Memphis MSA residents to assess their awareness of charitable organizations, tax code provisions for charitable donations, attitudes toward charitable organizations, and intentions to contribute in the coming year. Not everyone contacted will be eligible to take part in the study as there is a screening question which asks if the potential respondent is the person in the household that is most likely to make decisions regarding the family's contributions to charity. Because Margaret plans on using the research information to help her set objectives and the budget for the coming year, she has instructed the research firm to ensure that the information is representative of the MSA. Bob Smith, of the research firm, is using a probability sample and he is taking safeguards to minimize non-sampling error. Today, Bob is working on factors that will affect the response rate. Because there will be a screening question, Bob knows that this will:
Measurement Scale
Measurement is determining a description or amount of some property of an object that is ofinterest to the researcher.Marketing researchers are measuring properties --- sometimes called characteristics, attributes,or qualities --- of objects.Properties are the specific features or characteristics of an object that can be used to distinguishIt from another object. There are three categories of consumer characteristics that are of interest to marketing researchers and that can be measured and collected through questionnaires. 1. Behavior (objective properties that are observable).2. State of being (objective properties that are physically verifiable, e.g., demographics).3. State of mind (subjective properties that cannot be observed, e.g., attitude, perception, image). Objective properties are observable and tangible. Subjective properties are unobservable and intangible, and they must be translated ontoa rating scale through the process of scale development. Types of Measures - Nominal Measures To better design questions and response formats to measure the subjective properties of an object, we should have a good understanding of types of measures.Nominal measures are defined as those that use only labels; that is, they possess only the characteristics of description.Nominal measures are used to only categorize the respondents (e.g., race, religion, gender), without any quantitative value. Types of Measures - Ordinal Measures Ordinal measures permit the researcher to rank order the respondents or their responses. Ordinal measures indicate only relative size differences among objects: greater than, less than,or equal to. Types of Measures - Scale Measures There are two types of scale measures: (1) ratio scale and (2) interval scale. Scale measures are those in which the distance between each level is known. Ratio scale measures are ones in which a true zero origin exists (e.g., time, dollars).Interval scale measures are rating scales for subjective properties where the distance is normally defined as one scale unit (i.e., assumed interval). Table: Order of values is known- Ordinal: Yes, Interval: Yes, Ratio: Yes Can quantify the difference between each value- Interval: Yes, Ratio: Yes Can add or subtract values- Interval: Yes, Ratio: Yes Can Multiply or divide values- Ratio: Yes Has "true zero"- Ratio: Yes
Data Format
Measurement means assigning descriptors (number or label) to consumer characteristics accordingto certain pre-specified rules. Fact one: we are trying to measure consumer characteristics (behavior, state of being, state of mind) using questionnaires. Fact two: through measurement, we obtain our dataset where close-ended questions are coded as numbers but they have different meanings. There are four properties of measurement: assignment, order, difference, and origin. With assignment, we can classify consumers into categories through measurement. E.g., every VT students has a Hokie ID. E.g., consumer segmentation, current customers vs. potential customers. With order, we can rank consumers in terms of having more or less of some characteristicsthrough measurement. E.g., VT students receive letter grades (A, B, C, ‧‧‧). E.g., heavy users vs. light users of cigarettes, customers who are extremely satisfied vs. satisfied vs. dissatisfied vs. extremely dissatisfied. With difference, we are able to quantify the difference in characteristics among consumers in astandardized way. E.g., point grade, students could be freshmen, sophomores, juniors, or seniors (the difference between two adjacent standings is the same, one year). E.g., customers' product rating, customer satisfaction index. With origin, we have a unique, meaningful starting point (usually it is zero which has objectivemeaning) in a set of scale points when measuring certain consumer characteristics. E.g., VT students have age, salary, weight, and height. E.g., Consumers' willingness to pay for the product, how often consumers dine at a restaurant. We can classify the data you collected from the survey into four data formats: nominal, ordinal,interval, and ratio.They are hierarchical, with the lowest data format being nominal data while the highest being ratiodata.The reason for such classification is to determine appropriate statistical analysis. Nominal data is the data that serves as a label or tag (unordered category, no implied criteria toorder) to identify given consumer characteristics.Nominal data has assignment property only.Categorical questions provide nominal data.Frequency, percentile, and mode are available statistical operations. Categorical questions that ask respondents to choose the category that best describes them provide nominal data.Other examples include demographic questions such as "what is your gender?" and "what is youroccupation?" Ordinal data refers to data where consumer characteristics can be put in natural, ordered categories.Ordinal data has assignment and order property.The difference between data values cannot be determined.One-way labeled scale and rank-related questions provide ordinal data.Frequency, percentile, mode, and median are available statistical operations. One-way labeled (ranking) scale gives you ordinal data. Sometimes, we are interested in the rank order of several brands/products/stores.In Qualtrics, this can be achieved by using "rank order" questions, which gives you ordinal data. When you break natural metric questions into categories, you get ordinal data for the followingreason. In the "what is your age?" question, all information you get is that consumers who choose "25-34"Are older than those who choose "18-24". In other words, all you get an order. Interval data refers to data that permits making meaningful statements about the differences that separate characteristics between two consumers.Interval data has assignment, order, and difference property.The starting point is artificial, has no universal meaning, and in most cases is not available Because synthetic scales are used.Evenly-spaced synthetic scales such as Likert scale, N-point scale, semantic differential scale,and Stapel scale provide interval data.Most statistical analyses are available for interval data. Likert scale is so well designed that we usually take the data it records as interval data. Other rating scales such as N-point scale, semantic differential scale, and Stapel scale also provide interval data because they are evenly spaced. Ratio data is the data format that has a "true" starting point (no misunderstanding for the masspopulation) and a meaningful ratio.Ratio data has assignment, order, difference, and origin property.In most cases, natural scale metric questions give you ratio data but not always (we get theordinal data when we break a natural scale into several categories).All statistical operations can be performed on ratio data. Researchers are responsible for ensuring the ethical conduct of research. Respect for persons: protect people's privacy and freedom of choice. Beneficence: protecting research participants from harm (e.g., confidentiality of data). Justice: ensuring that participants are fairly compensated for their participation. We discuss the guidelines in terms of three phases of a research project: planning the research,collecting data, and actions following data collection. Ethical considerations while planning research: (1) voluntary participation and (2) informed consentto participate in the research. Voluntary participation means that participants' autonomy should be respected by giving them the choice of whether to participate.At the same time, the freedom to withdraw from the research should be guaranteed without Penalty once it has begun.For participation in research to be truly voluntary, potential participants must understand whatWill happen to them during the research. Ethical considerations during data collection: treating participants with courtesy and respect.The vast majority of research participants receive no monetary compensation.Their kindness should be reciprocated by kindness on our part.The most important part is the we should honor retractions of consent during the data collectionphase. Ethical considerations following data collection: (1) alleviating adverse effect, (2) educating participants about the study, and (3) maintaining the confidentiality of participants' data.Researchers must ensure that any adverse effects are removed.At the end of a study, participants are told about the purpose of the study. Data confidentiality involved protecting the data people provide from unauthorized disclosureto safeguard their right to privacy.
Both A and C (which are: "Don't you see problems with using your credit card for an online purchase?" AND "Were you satisfied with the restaurant's food and service?")
Which of the following questions is NOT an example of a loaded question?
Sampling 2
Non-probability (non-scientific) sampling is a sampling procedure by which the probability of each element(respondent)being selected in the sample cannot be known. We cannot guarantee that the non-probability sample is representative of the population so we cannot infer the population info from the sample because there exists non-sampling error. The advantage of the non-probability sample is low cost. Four popular non-probability sampling techniques are: 1) Convenience sampling 2) Purposive sampling 3) Chain referral sampling 4) Quota sampling Convenience (haphazard) sampling is a non-probability sampling procedure for obtaining respondents that are convenient to the researcher. Advantages: convenient (good for exploratory but bad for descriptive). Disadvantages: have non-sampling bias for sure. Convenience sampling, although fast and inexpensive, is often quite questionable when it comesto projecting the research findings. One reason for this is that convenient and cooperative units of the universe population often differ substantially from those units that are not quite soconvenient or cooperative. Purposive (judgment) sampling is non-probability sampling procedure based on judgment aboutsome appropriate characteristics required of the sample by experienced marketing researchers or experts. E.g., Focus group studies often use purposive sampling (when estimating the likely demand for low-fat, nutritious snacks, companies recruit participants based on their prior knowledge) Advantages: with valid judgment, the sample will do better than a convenience sample. Disadvantages: still have non-sampling error: with invalid judgment, it could go even worse than convenient sampling. With exit poll, polling companies uses a combination of convenience sampling and purposivesampling since polled voters who are selected to interview is based on both subjective judgment and convenience (voters often refuse to answer). During the Presidential Election in 2004, most Americans sat down to watch election night coverageexpecting that John Kerry had been elected President. The reason is that exit polls showed himahead in nearly every battleground state, in many cases by sizable margins. But the actual resultshows the opposite. What happened? Chain referral (snowball) sampling is a non-probability sampling procedure where existingrespondents recruit future respondents from among their acquaintances. Advantages: convenient and increased sample size. disadvantages: still have non-sampling bias. In the group project, chain referral sampling can be employed by asking family members or friends to distribute the survey on online social media. This can expedite the sample collectionprocess, when compared with convenience sampling. Quota sampling is a non-probability sampling procedure based on matching the populationon the specified control characteristics. E.g., knowing that audience of a TV program consists of 60% males and 30% Tivo users, researchers will recruit 100 respondents with 60 males and 40 females; 30 who have Tivo and 70 who do not. Advantages: with valid match, it greatly outperforms convenience and purposive sampling. Disadvantages: Not all characteristics are controlled for so there is still non-sampling error. sometimes it is difficult to find a sufficient number of desired respondents for control characteristics. Some demographic variables such as gender and age are best control characteristics for marketingresearchers. Probability (scientific) sampling is a sampling procedure by which the probability of each elementbeing included into the sample is known (it is equal in most cases). Probability sample is scientific so there is NO non-sampling error. By quantifying the sampling error (a confidence interval), we can draw inference on the population information using a probability sample. The downside is that it is very costly. Four popular non-probability sampling techniques are: Simple random sampling Systematic sampling Stratified sampling Cluster sampling. Simple random sampling is a probability sampling procedure in which each respondent is randomly selected and thus has equal probability of being included in the sample. Advantage: it has the strongest inferences to the general population. Disadvantage: it is expensive. Sometimes it is hard to obtain a complete sample frame. The old-fashioned way is as follows. Members of a population are numbered and numbers are put into a hat. Researchers randomly pick numbers from the hat. In this way, the population hasan equal chance of being selected. Nowadays, computers can help us generate random numberseasily. Systematic sampling is a probability sampling procedure that selects a random starting point andthen respondents are selected based on a fixed skip interval. Two key steps to guarantee success: Randomly select a starting point. Skip interval = round up (population size/sample size). Advantage: less expensive and less time consuming than SRS. Disadvantage: randomness may not be guaranteed. When conducting in-home survey, we usually use systematic sampling when selecting the household to visit. Select four houses from 11 households on a street. So, skip interval = round-up (11/4) = 3. Stratified sampling is probability sampling procedure in which (1) the population is divided intomutually exclusive and exhaustive subsets (called strata); (2) all strata are selected and elementswithin each stratum are randomly sampled. The premise of the stratified sampling:The number of strata is small. Consumers behave similarly within stratum and differently across strata. Advantage: it performs better than simple random sampling when the above conditions hold. The key to use stratified sampling is that sophomore, junior, and senior do not look like one another. Thus, we need to include some within each group into the sample Nielsen maintains a list of retailers where to collect sales info of consumer-packaged goods.However, retailers are quite different in size and market power. Overall, we can classify retailersinto four categories, ranging from large chain stores to small independent retailers. Thus, stratifiedsampling can be adopted by selecting a few retailers from each category into the sample. Cluster sampling is a probability sampling procedure in which (1) population is divided intomutually exclusive and exhaustive subsets (called cluster); (2) clusters are randomly sampledand then elements within each clusters are also randomly sampled. The premise of using cluster sampling is There are large number of clusters in the population. Consumers behave differently within a cluster and similarly across clusters. Advantage: it performs better than simple random sampling when the above conditions hold. Nielsen uses cluster sampling when determining the TV program audience rating by select a fewmetropolitan statistical areas (MSA) in U.S., a few cities in each selected MSA, a few zip codes in each selected city, a few street blocks in each selected zip code, and a few households in each selected street block. This type of cluster sampling has multiple layers of clusters. The choice between probability and non-probability sampling is a tradeoff between cost and the accuracy of information obtained.
5 Synthetic Scales
One-way labeled (ranking) scale is a label scale that is used to measure consumers' state of mindfrom one extreme to the other extreme. (a) one-way labeled satisfaction scale. Q. How would you rate the quality of care you received at Mt. Sinai Hospital? Poor₀Fair₀Good₀Very Good₀Excellent₀ (b) one-way labeled importance scale. Q. How important are the following items in your selection of a fastfood restaurant? Not at all Important₀Very unimportant₀Neither Important Nor Unimportant₀Very Important₀Extremely Important₀ (c) one-way labeled preference scale. Q. In thinking about your purchases of Ice Cream Brands, how strongly do you prefer any single brand?No Preference₀Slightly Prefer₀Prefer₀Strongly Prefer₀Very Strongly Prefer₀ (d) one-way labeled liking scale. Q. How was our service today? Didn't like at all₀Liked a little bit₀So-so₀Liked Somewhat₀Liked a whole lot₀ (e) one-way labeled likelihood scale. Q. How likely are you to recommend this brand to a friend? Very unlikely₀Unlikely₀Undecided₀Likely₀Very likely₀ N-Point (rating) scale is a number scale that is used to measure consumers' state of mind. We usually restrict the use of numbers to the positive side and begin with 1 as the smallest number. Q. How confident are you right now that your product choice is correct? Not at allConfident<1>₀<2>₀<3>₀<4>₀Very much Confident<5>₀ We usually anchor the largest and smallest numbers to some meaningful labels, which is calledanchored N-Point scale. Otherwise, it is called unanchored N-point scale. One-way Labeled: Advantage- The question is straightforward and does not require much work on the respondent side. Disadvantage- As researchers, we need to come up with these labels, which could be time-consuming. It has ordinal properties only. N-Point: Advantage- Number scale allows for more statistical analyses. Disadvantage- It requires the respondents to transform attitude, opinion, or intentions to number, leading to error and lots of works. Semantic differential scale- contains bipolar adjectives for various features of the brand/store/product and respondents choose their location based on their judgments. It translates qualitative judgments of brand/store/product into quantitative estimates. The focus of the semantic differential is on the measurement of the meaning of an object,concept, person, or experience. The ideal firm/product/brand is often used as the benchmark when the goal is improvement rather than comparison.By comparing your firm/product/brand with the ideal one, you obtain useful information on improvement. Semantic differential scale is widely used for brand profile analysis.However, only a few brands can be depicted. Plus, it does not provide information on the importanceof attributes.The value of semantic differential scale lies in its ability to draw quick comparisons. Q. How would you describe Kmart, Walmart, and Target on the following scale: 1____.2____3____4____5____Clean Dirty__________ __________Bright Dark___________ _________Low Quality High Quality________ ____________Conservative Innovative_________ ___________Inconvenient Convenient Stapel scale is a unipolar rating scale that measures respondents' feelings towards something ranged from a negative end to a positive end. Stapel scale can be easily recognized as it has numbers that range from a minus end to acorresponding plus end, with or without a zero as the midpoint. Stapel scale can also be used for brand/store/product image Q. Rate Folgers Decaffeinated Coffee on.. .+3+2+10-1-2-3Taste+3+2+1-1-2-3 Mild on Stomach Semantic Differential Scale: Advantage- No negative number involved.It is easy to compare. Disadvantage- It is difficult to construct bi-polar adjectives. Stapel Scale: Advantage- Unipolar adjective is sufficient.It is easier to construct. Disadvantage- Negative number involved. Likert scale asks the respondents to indicate the degree of (dis)agreement on a series of statements.Likert scale is best at capturing consumers' perceptions, value, attitude, and lifestyle from the intensity of (dis)agreement. Q. I generally buy the same brands I have always bought. StronglyDisagree₀Disagree₀SomewhatDisagree₀Neither AgreeNor Disagree₀Somewhat Agree₀Agree₀Strongly Agree₀ Issues with Interval Scales Used in Marketing Research 1. Do we need to include the middle, neutral response option? *A neutral option can give the opportunity to respondents to indicate ambivalence, but it can also be used as a method of hiding their opinions. 2. Do we need to use a symmetric or a non symmetric scale? *Some constructs (e.g., very few respondents will use the negative side as their responses)may require us to include only the positive side. 3. Do we need to include the negatively-keyed items? *It can be used to reduce agreement bias, but its performance can also be dropped. General Recommendations 1. Use precise descriptions for each category. 2. Use a balanced (i.e., symmetric) scale unless it is known that the respondents' attitudes are unbalanced (e.g., all favorable). 3. Use the odd number of categories if respondents could feel neutral, even if this is unlikely. 4. Use negatively-worded items only when it is necessary.
conditional branching/filtering question.
Patrick Strubel is a young attorney who is reviewing a proposed research survey to be used in an upcoming trial. The case involves potential deceptive advertising claims made in a TV commercial advertising over-the-counter (i.e., nonprescription) medicine by the defendant firm. Patrick is commissioning research to determine consumers' understanding of the claims after they have watched the TV commercial. Patrick wants to separate respondents into two categories: a. those who recall having actually seen the TV commercial and b. those who do not recall having seen it. For those who do not recall having seen the commercial, they will be asked to view the commercial before asking questions about their beliefs of the claims made. Patrick views the questionnaire proposed by the research firm. One of the questions is: 3. Do you recall having watched a TV commercial featuring XYZ, a pain reliever for arthritis? If "yes," continue. If "no," go to Question 12. This question would be considered a(n):
Sampling 1
Population is the aggregate of all elements from which information is needed.Elements could be customers, households, retailers, firms, sales territories, etc.The number of elements in the population is population size.As marketing researchers, we are interested in population information only. Census is the investigation of all elements that make up a population.Census is very costly and time-consuming to conduct.In many cases, it is almost impossible to conduct a census. This highlights the importance ofsampling and statistical methods. Sample is a subset of elements that make up the population.The number of elements in the sample is sample size, which is smaller than population size.When a sample is used instead of a population, population information is NEVER fully observedby marketing researchers. Sample is a subset of elements that make up the population. Sample information itself is of LITTLE interest to marketing researchers.Instead, we infer population information from sample information using hypothesis testing. Sampling is the process of obtaining a sample from the population. Whenever we use a sample rather than population, we need to worry about two types of errors Associated with inference from sample information to population information: 1) Sampling error 2) Non-sampling error Sampling error is the discrepancy between population info and sample info that arises when a representative sample is used instead of the population.Sampling error cannot be avoided or eliminated when a sample is used. Sampling error is the discrepancy between population info and sample info that arises whenonly a part of the population is included in the sampleThe sampling error strictly decreases with the sample size. When sample size is equal to population size, sampling error goes to zero. The sampling error (sampling error = population info - sample info) itself can be inferred usingstatistical analysis, but this requires more assumptions during the hypothesis testing process. Non-sampling error is the discrepancy between sample info from a sample that is representative of the population and that from an unrepresentative sample of equal size.Non-sampling error leads to bias in statistical inference, which renders it impossible to correctly infer the population information. Neither statistical procedures nor increasing the sample size can reduce or eliminate non-samplingerror.We need to design appropriate sampling technique to avoid non-sampling error. We witnessed another classic case when the non-sampling error biased the presidential electionIn 2016. The sample plan is a definite sequence of steps which the researcher goes through to draw andultimately arrive at the final sample. In this course, we propose the six-step sample plan process. Step 1: Define the target population. The target population is closely tied to the research objectives because it is defined early onwhen you determine the research objectives. Research objective (step 3 in the 11-step marketing research process) is what the researchermust do to solve the pre-defined marketing problem. It specifies... *What information is needed *From whom information is gathered => Target population *The unit of measurement used to gather the information *Word questions used to gather the information Step 2: Select a sample frame A sample frame is the source material or device from which a sample is drawn. It is a list of all those elements within a population that can be accessed. Some common sampling frames being used: *Mail address/email list maintained by a third-party survey company; *Customer list from companies' transaction history or loyalty program; *Consumers show up themselves (go to the store or browse the website); *Online sample frame maintained by online survey companies such as Qualtrics and onlineplatforms such as Amazon MTurk;Phone directory, yellow book. A sample frame is the source material or device from which a sample is drawn. E.g., technology survey over PAM students -> target population: all PAM students -> Sample frame: a list of PAM students from PAM undergraduate office. E.g., student satisfaction over marketing major -> target population: all marketing students ->sample frame: a list of marketing-majored students from the marketing office. Step 3: Determine Sampling Technique GRAPH Sampling Technique (at top)- connects to: 1)Non-Probability Sampling (non-scientific): Cheap and easy but likely to have non-sampling error *Convenience Sampling *Purposive Sampling *Chain referral Sampling *Quota Sampling 2) Probability Sampling (scientific): No non-sampling error, but expensive and difficult to attain *Simple Random Sampling *Systematic Sampling *Stratified Sampling *Cluster Sampling Step 4: Determine the sample size Methods for determining the sample size. Blind guess. Available budget. Standards for comparable studies. Rule of thumb: main group (n ≥ 100), subgroups (100 ≥ n ≥40) Statistical precision. Variability of population and individual strata. Acceptable level of sampling error. Needed level of confidence. Type of distribution. From the left to the right, we lose population information because more and more elements in the population becomes inaccessible.
Questionnaire design 3
Questionnaire organization concerns the sequence of questions that make up a questionnaireand the arrangement of the questionnaire.Questionnaire organization greatly affects the quality of information that is gathered. Three aspects of the questionnaire organization we focus on: 1) Logic (screening and filtering questions) 2) Introduction (included in the email or the opening of the questionnaire) 3) Flow (the sequence of questions) Screening questions are questions that are used to filter out unqualified respondents.In Qualtrics, screening is achieved through "skip logic" (i.e., skip towards the end of survey). A filtering (conditional branching) question is a question that determines what the next question(s) respondents see depending on how they answer it.It creates a custom path throughout the questionnaire that dynamically varies based onthe respondents' answers. When using multiple filtering questions, it should be noted that respondents will get savvyto the procedure and intentionally provide answers to the filter questions that will not triggerthe follow-up questions. In Qualtrics, "display logic" and "branch logic" can be used to achieve this goal. 6 Functions accomplished by Introduction: Function 1 --- Identify the surveyor. Ex: Hello, my name is XXX, and I am a telephone interviewer working withNationwide Opinion Research company here in Milwaukee. I am notselling anything. The sponsor of the survey is divulged. Plus, the prospective respondent is madeaware that this is a bona fide survey and not a sales pitch. Function 2 --- Indicate the purpose of the survey Ex: We are conducting a survey on Internet browsers, sometimes called"Web browsers." It informs prospective respondents of the topic and the reason for the survey. Function 3 --- Explain how the respondent was selected. Ex: Your telephone number was generated randomly by a computer. It notifies prospective respondent how s/he was chosen to be in the survey. Function 4 --- Requests for incentive for participation. Ex: This is an anonymous survey, and I would like to ask you a few questionsabout your experiences with your Web browser program. It asks for prospective respondents' agreement to take part in the survey.Note that anonymity helps you gain cooperation. Function 5 --- Screen out unqualified respondents. Ex: Do you use Internet Explorer, Firefox, or Opera? It determines if prospective respondents are qualified to take part in the survey. Those whodo not use one of these programs will be screened out. Function 6 --- Express appreciation. Ex: Your opinions are important as they will help us (1) better understandour customers, (2) improve our website design, and (3) complete this study. We truly appreciate your cooperation and assistance. If there is no monetary incentive, then tell them their responses will be used to improvesomething like customer service or product, which will benefit them. The funnel approach of question flow to decrease drop-out rate. From easy to difficult questions, from general to specific and then to sensitive questions. 1. Screens (first questions asked). Ex: "Have you shopped at Old Navy in the past months?""Is this your first visit to this store?" Screen questions are used to select the respondent types desired by the researcher to bein the survey. 2. Warm-up behavioral questions (immediately after any screens). Ex: "How often do you go shopping for casual clothes?""On what days of week do you usually shop for casual clothes?" Easy to answer questions; shows respondents that survey is easy to complete; generatesinterest. 3. Transitions (prior to major sections of questions or changes in question format). Ex: "Now, for the next few questions, I want to ask about your family's television viewing habits.""Next, I am going to read several statements and, after each, I want you to tell me if you agree or disagree with this statement." Notifies respondents that the subject or format of the following questions will change. 4. Complicated and difficult-to-answer questions (in the middle of the questionnaire). Ex: "Rate each of the following ten stores on the friendliness of their sales- people on a scale of 1 to 7.""How likely are you to purchase each of the following items in the next three months?" Respondent is committed to completing the questionnaire; can see (or is told) that there arenot many questions left. 5. Classification and demographic questions (in the last section). Ex: "What is the highest level of education you have attained?""What is your age?""What is your gender?" Questions that are "personal" and possibly offensive are placed at the end of the questionnaire.
all of the above (which are: affect the quality of the information gathered. ; motivate respondents to be conscientious. ; discourage respondents. ; cause respondents to stop answering questions.)
Questionnaire organization has the potential to:
none of the above: (which are: A table of answers and brief statistics regarding student answer choices. Answer Text Number of Respondents Percent of respondents selecting this answer Answer Distribution the order of words within a question., (Incorrect answer)the order of words within a question. ; the order in which a firm conducts questionnaires. ; the ability of a researcher to produce an organized questionnaire. ; a professional group of experts whose specialty is designing questionnaires.)
Questionnaire organization refers to:
Questionnaire design 2
Questionnaire wording: When developing a questionnaire, keep in mind that a single word can make a difference in howparticipants respond to a question.For example, participants of a study reported more "Don't know" when they were asked "did you see the broken headlight?" than "did you see a broken headlight?"That is, as little as one word in a question can result in question bias that will distort survey findings. Question evaluation amounts to scrutinizing the wording of a question to ensure that questionbias is minimized and that the question is worded so that respondents understand it and canrespond to it with relative ease.To achieve this goal, researchers need to get into a respondent state of mind.Some tips are offered to achieve this goal: four do's and six do-not's. Four "Do's" in Questionnaire Wording: 1) The question should be focused on a single issue or topic. Questions that violate this guideline are vague and prone to inconsistent interpretation. e.g., How do you feel about Walmart? Correct: Please rate each of the following aspects of Walmart on a 5-point scale with 1 being the worst and 5 being the best. e.g., What type of hotel do you usually stay in when on a trip? Correct: When you are on a family vacation, what type of destination hotel do you typically use? 2) The question should be brief (remove redundant words). Unnecessary and redundant words should always be eliminated. Brevity of your questionnaire helps the respondent comprehend the central question and reduceThe distraction of wordiness. e.g., When some appliance in your home breaks, will it come into your mind that you can call Sears repair service to come and fix it? Correct: When you are in need, do you call Sears repair service? Eliminate wordy and redundant expressions. 3) Questions should be grammatically simple (has only a single subject and predicate). A simple sentence is preferred because compound and complex sentences are busy with multiple subjects, predicates, objects, and complements. The more complex the sentence, the greater the potential for respondent error. e.g., If the Sears repair service schedule was not convenient for you, would you consider or not consider calling a competing repair organization to fix the problem you have? Correct: If you did not use Sears repair service, would you use another repair service? 4) Questions should be crystal clear.You should avoid words that are vague or open to misperception and use words that are in respondents' core vocabularies.It should be achieved even at the expense of briefness and simplicity. e.g., How many children do you have? Correct: How many children under the age of 18 living with you in your home? Six "Do-not's" in Questionnaire Wording: 1) Avoid ambiguous words in questions. Do not use words of extreme absolutes, meaning that respondents are placed in an extreme position where they must either completely agree or disagree. Be careful when using grading adverbs and gradable adjectives in Yes-or-No dichotomous categorical questions. e.g., Do you shop online regulary? Correct: How often do you go shopping online per month? 1, 2, 3-5, 6+ e.g., Do you think this Campbell soup is hot? Correct: On a 7-point scale, how hot is this Campbell soup? 2) Avoid leading questions that give respondents a strong (expectation) or subtle cue as to how to answer. Extensive use of leading questions can lead to the generation of false memories and hencebiased answers (think about interrogation).Wording difference can alter participants' responses. For instance, strong verbs, such as "force"and "prohibit" represent control or action and can lead to biased response. e.g., do you agree that the government should force you to pay higher taxes? Correct: Use less stronger words such as "require" and "mandate." e.g., How would you rate the fantastic performance of legendary Joe Brady? Correct: How would you rate the performance of Joe Brady? Include socially desirable outcome (subtle cue). e.g., Have you dined at a newly opened fast-food restaurant XYZ that everyone is talking about? Correct: Have you dined at a newly opened fast-food restaurant XYZ? Use preceding questions to set up assumptions in the mind of the respondents (subtle cue). e.g., Since you agreed that fast-food consumption leads to eating disorders, should fast-food firms be responsible for correcting them? Correct: Should fast-food firms be responsible for correcting eating disorders? Include answer or apparent logic in the question (strong cue). e.g., Don't you think that fast foods have too many calories? Correct: Do you think that fast foods have too many calories? e.g., You like fast food, don't you? Correct: Do you like fast food? e.g., Since most fast foods are fried, should these companies put warnings on their labels? Correct: Should these companies put warnings about fried food on their labels? 3) Avoid overstated questions that place undue emphasis on some aspects of the topic.When the overstated part leads respondents to choose a particular answer, an overstated questioncan also be a leading question. Overstate supported facts in an unbalanced fashion (leading). e.g., As studies have shown that overweight children buy fast foods, should "kids meals" be healthier? Correct: Do you think "kids meals" should be healthier? e.g., How much would you pay for sunglasses that will protect your eyes from the sun's harmful ultraviolet rays, which are known to cause blindness? Correct: How much would you pay for sunglasses that will protect your eyes from the sun's harmful ultraviolet rays? Overstate wrong assertions or facts. e.g., As everyone buys fast foods, should everyone be concerned about nutrition? Correct: Are you concerned about the nutrition of fast food? Overstate answers to preceding questions to set up assumptions in the mind of the respondents(likely to be leading). e.g., Since you agreed that fast-food consumption leads to eating disorders, should fast- food firms be responsible for correcting them? Correct: Should fast-food firms be responsible for correcting them? 4) Avoid loaded questions that make a reference to facts, universal beliefs, or unjustified assumptions that generate an emotional/cognitive impetus toward an answer.When the loaded facts or universal beliefs are leading, a loaded question can also be a leadingquestion.When the loaded facts or universal beliefs are overstated, a loaded question can also be anoverstated question.Questions loaded with unjustified assumptions e.g., Did you stop smoking? Correct: Do you smoke? Did you stop smoking? Questions loaded with facts. e.g., Do you think that KFC should have warning labels for obese people suffering fromdeadly diseases such as diabetes? Correct: Do you think that KFC should have warning labels for obese people? Questions loaded with general beliefs. e.g., Should people be allowed to protect themselves from harm by using taser guns as self-defense? Correct: Should people be allowed to use taser guns as self-defense? 5) Avoid using double-barreled questions that have two questions in one.With two questions posed together, it is difficult for respondents to answer either one directly,thus, providing misleading information. e.g., Are you satisfied with KFC's food and services? Correct: Are you satisfied with KFC's food? Are you satisfied with KFC's services? 6) Avoid ill-defined questions. Details they don't recall. e.g., How many and what brands of aspirin did you see the last time you bought aspirin? Correct: When was your first time to buy aspirin? Questions that invite guesses. e.g., What is per-gallon price of premium gasoline at the Exxon station on the corner? To predict their actions in circumstances they cannot perform. e.g., How often would you go out to eat at this new, upscale restaurant that will be built 10 miles from your home?
Nominal data
Race, religion, type of dwelling, gender, brand last purchased, and buyer/nonbuyer are examples of:
A and B. (which is: costs. AND not all elements in the population are accessible.)
Reasons for taking a sample instead of a census include:
One-way labeled scale.
Which of the following scales might have an example that ranges from "Not Important" to "Extremely Important"?
subjective properties
Researchers often desire to measure ________, which cannot be directly observed because they are mental constructs such as a person's attitude or intentions.
ordinal data
Rick Irizarry of Irizarry Insurance Company is conducting a survey of potential customers. One of his questions asks how much customers would be willing to pay in annual premium for a $100,000 life insurance policy in the following way: Check the appropriate category: ___ $1 to $10 month ___ $11 to $20 month ___ $21 to $30 month ___ $31 to $40 month ___ $41 to $50 month
synthetic metric
Scale development is primarily concerned with the creation or use of ________ measures.
determine whether or not a potential respondent qualifies to take part in a survey based on certain selection criteria the researcher has deemed essential.
Screening questions are used to:
Questionnaire design 1
Seven Tips Tip #1: Keep the survey not too long or short. But, make it simple and clear!4~5 mins is a good range for most questionnaires (especially for online survey sent to collegestudents).Long survey could lead respondents to drop out (more pronounced in online survey) and tostop paying attention. Tip #2: Use scales whenever possible!You will get more information from each question.Scales can measure both the direction and intensity of opinions. When using scales, limit the number of scale levels to 5 or 7. Tip #3: Always explain why! Explain unexpected questions.e.g., When asking a toy store consumer about online shopping behavior, you should explain why or bridge the gap.Justify requests for sensitive information.e.g., Your purchasing habits will only be analyzed in aggregate for benchmarking purposes and they will NOT be shared with any firm. Tip #4: Speak your respondents' language! Do NOT use industry specific jargon and acronyms.e.g., caloric content, bits, bytes, SWOT, etc.The best move is to write to your least-informed respondent. Tip #5: Closed-ended questions require attention!Give collectively exhaustive and mutually exclusive choices.If not, use "Other (specify: ___________)" e.g., What is your age? 1) 0-10, 2) 10-20, 3) 20-30, 4) 30-40, 5) 40+ Limit the number of alternatives to reduce cognitive bias. The explicit alternatives listed should cover 95% of respondents.Add a "Prefer not to answer" option whenever necessary.e.g., What is your race? What are your religious beliefs? What is your age? Tip #6: Utilize Qualtrics to make the questionnaire more interactive and user-friendly!Add a progress bar (keep respondents posted about the progress). Add "Force response" option when necessary.Explore more advanced features. Tip #7: Take your survey for a test drive!It is difficult to avoid last little issues like occasional typo, misdirected question, or unfamiliarbuzz word. How long did it take? (be sure to keep it short).Find at least 5 people from your target respondents to take your survey and askOverall impression of the questionnaire.Which questions were confusing?Any other suggestions or problems during the survey? Questionnaire design is a systematic process that requires the researcher to go through a series of considerations. 1)Researchers contemplate various question formats. 2)Researchers word the various questions carefully. 3)Researchers organize the layout of the questionnaire. 1 and 2 are Question Development Open-ended questions instruct the respondents to respond in their own words without anyoptions provided.They are usually used (1) when there are too many potential options (unaided) and (2) to ask for additional information or clarification (aided).They should be used with caution (1 or 2 are enough) since they are NOT standardized questions. Open-ended questions is often used to ask for comments and feedbacks. Aided open-ended question is usually placed after the unaided open-ended question to ask for further clarification or further information. Use open-ended questions with caution. Advantages of open-ended questions.Avoid bias in offering explicit choices (no exhaustive and mutually exclusive options can be provided).Allow respondents some degree of control. Disadvantages of open-ended questions.Coding of open-ended question is time and effort consuming.Idiosyncratic use of language ("yes, I use it sometimes").Giving extra weight to respondents who are more articulate. Categorical questions ask the respondents to choose the category that best describes them.The major difference with metric question is that, to answer categorical questions, we doNOT need a scale or measurement on the scale.Categorical questions can be either dichotomous or dual-choice (with two options) orMultiple-choice (with more than two options).Categorical questions provide nominal data. Metric questions call for measurement on a scale by the respondents.Metric questions ask for the exact level of a variable and thus provide ordinal, interval, and ratio data. Options of metric questions can be: what is your age? 18-24, 24-30 etc. Advantages of closed-ended questions. Allow simple coding and numerical analysis. Allow uniformity of treatment and specific research design. Quicker, easier for respondents. Disadvantages of closed-ended questions.A premise is that we must know what are possible options beforehand.Respondents are less involved in their answers.
it allows respondents to dodge or hide their feelings.
Some researchers believe that a mid-point, or neutral position, should NOT be included in a scale question because: Hint: Thus, sometimes, marketing researchers choose not to use a neutral point intentionally to force respondents to pick a side.
It is suitable for mail, telephone, or personal interviews.
The Likert scale has several advantages. Which of the following is not an advantage?
population
The ________ is the entire group under study as specified by a research project.
define the target population.
The first step in developing a sampling plan is to:
B and C. (which are: The alternatives are not exhaustive. AND The categories are not mutually exclusive.)
What is wrong with this question?What is your annual income? (please check)__$10,000-$25,000__$25,000-$40,000__$40,000-$50,000__$50,000-$75,000__$75,000-$100,000
individuals responsible for cleaning service in companies located in Iowa.
The population for a market research project in Iowa to determine the cleaning service needs in companies would be defined as:
The question should be a grammatically simple sentence if possible.
The question ""If you were looking for an automobile that would be used by the head of your household who is primarily responsible for driving your children to and from school, music lessons, and friends' houses, how much would you and your spouse discuss the safety features of one of the cars you took for a test drive?" violates which of the following rules of question wording?
leading question
The question "Don't you see any problem with using your credit card online?" is a:
is ambiguous
The question, "Do you exercise often? seldom? sometimes?" __________.
A and C (which are: is leading AND is double-barreled)
The question, "Do you feel the government should be forced to stop picking our pockets with excessive taxes and return a portion of the taxes collected to taxpayers?" _________.
is double-barreled
The question, "How do you judge the price and the quality of the product?" _________.
a leading question.
The question: "Because the Founding Fathers of the United States provided citizens the right to keep and bear arms in the Bill of Rights, shouldn't you be against gun control legislation?" is:
Unaided open-ended.
The question: "What was your reaction to the Sony CD player advertisement you saw on television last?" is an example of which response format?
A table of answers and brief statistics regarding student answer choices. Answer Text Number of Respondents Percent of respondents selecting this answer Answer Distribution sample error decreases., (Correct answer)sample error decreases.
The relationship between sample size and sample accuracy is that as sample size increases:
Pretest participants should be representative of the target population under study.
There are several parameters regarding the pretest of a questionnaire. Of the following, which is MOST accurate?
Open-end, categorical and metric.
What are the three basic question formats?
Interval.
You have been assigned the task of evaluating consumer acceptance of and favorability toward a new wedge that Stuart Weitzman (a high-end shoe company) has just test marketed in your area. In order to determine the degree to which consumers hold favorable attitudes toward the shoe, what is the lowest data format that you can use?
Incentives; increase
________ are offers to do something for the respondents in order to ________ the probability that the respondents will participate in the survey.
Objective Properties
________ are physically verifiable characteristics such as age, income, number of bottles purchased, store last visited, and so on.
Sample frame
________ is the list of elements in the population from which the sample is selected.
Pretesting
________ is the testing of the questionnaire on a small sample of respondents for the purpose of improving the questionnaire by identifying and eliminating potential problems.
Screening
________ questions are used to ferret out respondents who do not meet certain qualifications.
both A, B and C. (which are: represent the true population as much as possible. ; have a reasonable cost. ; is accessible)
t is the researcher's responsibility to select only sample frames that: