790 likes | 1.13k Views
University of Washington EMBA Program Regional 20. Marketing Management “Marketing Research” Instructor: Elizabeth Stearns With Survey review from Professor Dan Turner. Marketing Research Agenda. Description and Overview Choosing to do Marketing Research
E N D
University of Washington EMBA ProgramRegional 20 Marketing Management “Marketing Research” Instructor: Elizabeth Stearns With Survey review from Professor Dan Turner
Marketing Research Agenda • Description and Overview • Choosing to do Marketing Research • Survey & Questionnaire Design • Conjoint Analysis • Advertising Research
Marketing Research • A disciplined approach to the development and provision of information relevant to marketing decision-making. • Research should be focused on addressing questions where different answers would alter the decision a rational manager makes
Fit the research to the problem! Open Blind Prefer Pepsi 51% PreferCoke 44% Equal/Can’t Say 5%
Understand your what your customers value. Open Blind Prefer Pepsi 51% PreferCoke 44% Equal/Can’t Say 5%
Management Decision Problem What should the decision maker do? Should the price of the brand be increased? Should we change the advertising campaign? Marketing Research Problem What information is needed and how can it be obtained? Two Layers of Decision Problems What is the price elasticity of demand? What is the anticipated impact on sales and profits of the price change? How effective is the current campaign in generating awareness?
Should Research be Conducted? Time Constraints Availability of Data Decision Importance Value vs. Costs Is there sufficient time available before a managerial decision must be made? Does the value of the research information exceed the cost of conducting research? Is the information already on hand inadequate for making the decision? Is the decision of considerable strategic or tactical importance? Conduct Marketing Research Y Y Y Y N N N N Marketing Research Should Not be Conducted
Backward Market Research • Start the process where it usually ends and then work backward • Determine how the research will be implemented • Determine what the final report should contain and how it should look • Begin with an end in mind • Close collaboration between the researcher and the organizational decision maker is the single most important factor in obtaining a desirable outcome
Marketing Research • Qualitative: Exploratory/Developmental • Help frame the problem • Provide insight and ideas • Generate testable hypotheses • Choose variables for inclusion in a data set • Add richness to interpretation of relationships that are uncovered through quantitative analysis • Quantitative: Descriptive/Evaluative • Detect & describe relationships between variables in data • Sophisticated quantitative analysis cannot compensate for framing the problem inappropriately and, hence, examining the wrong set of data. Nor can it insure the correct interpretation of a relationship that is uncovered. • Causal: Experimental
Choosing Research Designs Sample Marketing Application Data Collection Examples Research Design • EXPLORATORY (Qualitative) • * Identifies motivations • underlying behavior • * Yields initial hunches & • insights for future research Opportunity Analysis Issue Identification • Focus Groups • Ethnographies • Depth Interviews • Projective Techniques • DESCRIPTIVE (Quantitative) • * Describes consumer behavior • Who, when, where, how much • * Verifies insights with • numerical data for action • Observational Studies • Surveys (one time) • Consumer Panels(longetudinal) • A & U Studies Target Profile Identification Demand Estimation • CAUSAL (Experimental) • *Identify consumer response • to marketing strategies • * Formally establishes • cause and effect relationship • Lab Experiments • Field Experiments • Quasi-Experiments Strategy Formulation Strategy Evaluation
Sources of Marketing Data PRIMARY DATA SOURCES SECONDARY DATA SOURCES External Sources Internal Sources External Sources Internal Sources • Market research firms • Advertising agencies • In-House marketing • research department • Government agencies • Competitors • Trade associations • Business publications • Syndicated sources • Store audits • UPC scanner data • Advtg. exposure data • Single source data • Specialized syndicated • products/expert systems • Accounting data • Sales reports • Factory shipments
Survey Basics • Background: Surveys as part of the marketing research efforts • Managing the Survey Effort • Item Development • Questionnaire Organization • Critical Reviews & Pretests • Additional Resources
Background—Surveys as Part of the Marketing Research Effort • Form of primary data collection for descriptive research • Can be either cross-sectional (snapshot/photograph) or longitudinal (movie) • Easier to obtain representativeness with cross-section but more difficult to detect changes over time • Surveys administered via interview (personal, phone) or “writing” (paper & pencil, web/email)
Key Questions in Survey Research • What managerial decision is to be made? • What information will best help us make the decision? Begin with an end in mind! • What survey questions/items can get us that information from respondents? • How should those questions/items be phrased? • Who should our respondents be? • How are we going to contact respondents? • How many should we get? • What will we revise based upon the pretest? • How do we analyze and draw inferences from the data?
Managing the Survey Effort : Survey Mode • Primary considerations are effectiveness, speed, monetary cost • Personal interview/intercept • Negatives—Expensive, respondent/interviewer biases • Positive—Flexibility • Telephone • Negatives—Requires very simple items, representativeness issues • Positive—Quick • Paper & pencil/Mail • Negative—Requires simple items • Positive—Respondent privacy/anonymity • Email/web • Negative—Sampling frame bias, requires simple items • Positives—Quick, cheap
Managing the Survey Effort : Sampling • Probability sampling—Every element has a known (or knowable) non-zero probability of being included in the sample • Non-probability sampling—Sampling without random selection • Increasing sample size reduces random sampling error but does nothing to reduce biases • To reduce sampling error you have to be rich, and to reduce sample bias you have to be smart
Managing the Survey Effort: Total Survey Error • Sampling error—Unless you measure the whole population (a census) your sample estimate will not likely equal the exact population value • “Margin of error” noted with survey results is typically an estimate of the sampling error • Nonsampling error • Nonresponse bias—Certain members of the original sample do not provide responses • Measurement error—Instruments used to gather observations produce systematic error
Managing the Survey Effort : Controlling Nonresponse Bias • Two basic determinants of who responds to surveys and who does not • Availability • Choose data collection method that is best suited to the respondents • Motivation • Some element is exogenous, e.g., BCC survey vs. canned soup survey for UW MBA students • Controllable factors • Minimize cost of responding • Maximize the rewards • Establish trust that rewards will be delivered
Managing the Survey Effort : Controlling Nonresponse Bias • Reward respondents • Show positive regard and give verbal appreciation • Support respondent values • Offer tangible rewards (including possibly sharing results/insights) • Write an interesting questionnaire • Reduce costs to respondents • Make the task appear brief • Reduce physical and mental effort requirements • Minimize the possibility for embarrassment • Eliminate direct and indirect monetary costs • Follow-up with nonrespondents in a second survey wave • Establish trust • Provide a token of appreciation in advance • Borrow legitimacy by establishing ties to a trusted organization
Managing the Survey Effort : Measurement Error • Reliability—Stability, consistency, or reproducibility of a measure • Validity—Measurement of the construct or concept intended • Examples • “Short” ruler to measure distance • GMAT to measure business school acumen • Insights regarding measurement error • The greater the degree of structure the lower the magnitude of measurement error • The greater the self-presentation elicited by the question the greater the risk of measurement error
Item Selection: Laws of Questionnaire Design • Information Test: Don’t ask a question unless truthful answers to it will provide useful information in making the decision at hand. • How (specifically) will I use the data from this question? • If the answer is no more precise than “I’ll analyze it” you probably don’t need that question. • If there is more than one way to get a particular piece of information, pick the items for which respondents are more likely to both… • know the answer AND • be willing to tell you the answer
Item Design: Types of Items • Closed-ended questions with categories • Semantic differential • Likert scale • Object ranking • Constant sum scale/Point Allocation • Open-ended • Numeric • Textual
Item Design: Types of Items • Closed-ended questions with categories What marathon training programs did you consider before choosing Team in Training? ___ Joints in Motion ___ Team Diabetes ___ Official Asics Training ___ Always Running ___ Chuckit ___ Team Danskin ___ Other ________________ (please specify) • Make sure categories are mutually exclusive and collectively exhaustive
Item Design: Closed-ended Purchase Intent Scale If a set of three ant traps sold for approximately $1.00 and was available in stores where you normally shop, would you: ___ Definitely buy the traps ___ Probably buy ___ Probably not buy ___ Definitely not buy
Item Design: Types of Items • Semantic Differential • Most popular way in marketing research to obtain attitude toward an object • Very useful for rating several alternatives on a given set of attributes • Anchored at each end by an adjective • Often bipolar (Sweet-Sour) • Sometimes monopolar (Sweet-Not Sweet) • Respondent marks the point which best describes the object(s) Old New
Item Design: Semantic Differential Scale Lexus IS300 and BMW 3 Series Unattractive --- --- --- --- --- --- --- AttractiveAppearance Appearance Noisy --- --- --- --- --- --- --- Quiet Reliable --- --- --- --- --- --- --- Unreliable Built well --- --- --- --- --- --- --- Built poorly Good value for --- --- --- --- --- --- --- Poor value for the money the money
Item Design: Likert Scale Strongly Strongly Disagree Agree They might make my feet feel hot 1 2 3 4 5 I am satisfied with what I am using 1 2 3 4 5 My problem is not serious enough 1 2 3 4 5 Too much trouble to cut them to fit 1 2 3 4 5 Price is too expensive 1 2 3 4 5 Might make my shoes too tight 1 2 3 4 5 I’m embarrassed to buy them 1 2 3 4 5
Item Design: Constant Sum Scale/Point Allocation Below are ten characteristics of women’s tennis sportswear. Please allocate 100 points among the characteristics such that the allocation represents the importance of each characteristic to you. The more points that you assign to a characteristic, the more important it is. If the characteristic is totally unimportant, you should no allocate any points to it. When you’ve finished, please double check to make sure that your total adds to 100. • Disadvantage: High level of respondent effort • Advantage: Ratio level measurement
Item Design: Point Allocation Example Characteristics of Tennis Sportswear Number of Points Is comfortable to wear _______ Is durable _______ Has the endorsement of a famous athlete _______ Is made by well-known brand or sports manufacturers _______ Is made in the U.S.A _______ Has up-to-date styling _______ Is flattering to the body _______ Gives freedom to the body _______ Is a good value for the money _______ Authentic, like the pros wear _______ _______ 100 Points
Item Design: Key Issues Checklist • Can the research objective be fulfilled without asking this question? • There is a cost to each item. • Do respondents have the information and motivation to answer each item? • Do not ask what respondents cannot or will not divulge. • Are the questions clear and unequivocal? Will the words be universally and uniformly understood? • For example, “How many members are there in your family?” • Does the question use a double negative? • “Are there no circumstances under which you would not use a stain removal detergent enhancer?”
Item Design: Key Issues Checklist • Are the questions leading/loaded? • Do you think the US should allow public speeches against democracy? • Do you think the US should forbid public speeches against democracy? • 44% of respondents replied “No” to the first question • while 28% of respondents in a similar sample said “Yes” to the second question
Item Design: Key Issues Checklist • Are the questions double-barreled? • Do you believe McDonald’s offers fast and courteous service? • Are responses mutually exclusive and collectively exhaustive? • Are the questions too complex? • Of the total number of miles you have driven during the past month, approximately what percentage was for driving to and from work? • Are the question implications implicit or explicit? • Are you in favor of UW Business School providing every student with a laptop? • …even though it will mean a substantial increase in tuition?
Questionnaire Organization • Introduction: Basic who, what, when, where, why, how We are a group of UW Executive Business School students working on a school-related project. We are conducting a marketing research survey to determine interest in a new child safety product. Your attitudes and opinions will be valuable in designing a child safety product that is best suited to satisfying market needs. We are not selling anything, and you will not be solicited later. Your responses will be used for research purposes only, and the input you provide will be both completely anonymous and confidential. We anticipate that this survey will take you less than 5 minutes to complete. THANKS IN ADVANCE FOR YOUR PARTICIPATION
Questionnaire Organization • Body—Theories of Organization • Typically begin with one or two interesting but non-taxing/non-invasive “warm-up” questions • In general order items in terms of descending usefulness or importance to you • Group questions that have similar content or similar form (e.g., Likert scale items) together • Leave potentially objectionable and sensitive items to the end of the survey • Thank respondents for participating
Questionnaire Organization • Body: Typical Item Sequence • Screeners/qualifying questions “Have you been skiing in the past 12 months?” • Warm-ups “What brand of skis do you own?” • Beginning body/transitions “What features do you look for in skis?” • Middle—more complicated/difficult questions “Please rate the importance of the following ski attributes in your ski purchase decisions.” • End—classifications, demographics, and sensitive items “What is the highest level of education you have obtained?”
Critical Reviews and Pretests • Always pretest among a small group of people similar to those future respondents of the actual survey • Get qualitative feedback on troublesome questions, ambiguities, etc. after the pretest group completes the survey If you don’t have the resources to pilot test your survey then don’t do the study. • For each item, ask “Is this question really necessary or merely interesting?” • Keep it short and simple
Conjoint Analysis • Technique to understand how consumers make trade-offs among attributes or characteristics of products or services which deliver the desired benefit. • To help management understand the implications of such trade-offs for design of product offerings • Conjoint Analysis also provides a measure of how important the attributes are to the customer • Conjoint Analysis is now quite widely used by marketing research companies, consulting companies, etc.
How Does It Work? Suppose I want to know what factors are important to you in your evaluation of cars. You can think of a car as a bundle of features or attributes e.g. size - small, mid-size, full size; fwd/rwd; mpg - 20, 25, 35; engine - 4 cl, 6 cl; American/Japanese/European; many dealers/few dealers; $15k/20k/25k; low interest financing or cash back, etc. Now how do you go about buying a car - you have an approximate budget, you want certain features, you would pay more for a fwd than a rwd, more for an automatic than manual, etc. Assume there is no optimum car or if there is one it is out of your budget. Some of the attributes listed above are not directly comparable. You like the prestige of a German car (say) but you may have to put up with the inconvenience of longer down-time if the car has to be serviced. You like fuel efficiency but you don't like the cramped feeling of a small car. And so on.
HowDoes It Work? If you could assign a $ value to every attribute - how much more would you pay for a bigger engine, for prestige, for fuel efficiency, for size, etc. -then you could compute the equivalent $ value of the cars in your set and make a choice. However, in many cases this is not possible. You may make some mental computations and come up with an overall value or utility preference for the cars in your set. This overall value is presumably based on component values assigned to the attributes.
How Does It Work? It is much easier for the consumer to provide overall preferences than the component utilities or values. Conjoint analysis takes these overall preferences and the attribute descriptions and decomposes the overall preferences into utilities for the attributes. The utilities of different attributes are comparable. These utility values provide the trade-offs among the attributes. From these values we can construct an importance measure for each attribute.
Managerial Uses of Conjoint • After determining the contribution of each attribute, the researcher could • Define the object with the optimum combination of features • Predict market shares among objects with different sets of features • Isolate groups of consumers who place differing importance on features • Identify market potential for product concepts not yet available
Commercial Applications • Commercial applications are across the board: • Consumer durables • Automobiles, refrigerators, car stereos, condominiums • Consumer non-durables • Bar soaps, hair shampoos, disposable diapers, clear gravy (!) • Industrial Products • Copy machines, forklift trucks, computer software, aircraft • Services • Car rental, credit cards, hotels, mass transit
Steps in Conjoint Analysis 1. Develop the set of attributes 2. Select the levels of each attribute 3. Develop the set of stimuli to be used 4. Obtain an evaluation (rating or ranking) of the stimuli from consumers 5. Estimate the utility values for each level of each attribute 6. Compute importance weights for each attribute 7. Evaluate the trade-offs among attributes and which combination of features is most preferred 8. Affiliate results across consumers 9. Conduct market simulation
Desirable Problem Situations for Conjoint • Product must be realistically decomposable into basic attributes. • Product choice tends to be a reasoned decision. • Factorial combinations of basic attribute levels are reasonable (i.e., some combinations are not unreasonable).
Example: Carpet Cleaner Levels A, B, C K2R, Glory, Bissell $1.19, $1.39, $1.59 Yes, No Yes, No Attributes Package Design Brand Name Price Good Housekeeping Seal Money-back Guarantee • Total number of possible product combinations = 108 • A subset of 18 product combinations is selected for use such • that attributes are uncorrelated (orthogonal). • Which 18? Commercial software available to pick subset. • Why 18? To allow sufficient degrees of freedom to estimate • the model for each individual respondent.
Brand name Retail Price -- -- -- -- -- -- -- -- -- -- 1.0 1.0 Utility Utility | | | K2R Glory Bissell | | | $1.19 $1.39 $1.59 0 0 Money-Back Guarantee? Good Housekeeping Seal? -- -- -- -- -- -- -- -- -- -- 1.0 1.0 Utility Utility 0 | | No Yes | | No Yes 0 Conjoint Analysis ©2000 Prentice Hall
Desirable Problem Situations for Conjoint • Product can be realistically described, verbally or pictorially. • Desirable new product alternatives can be synthesized from basic attributes.