Avoid these four mistakes in marketing research
- 1. Types of Marketing Research & Common Errors in Marketing Research Vikas Mittal J. Hugh Liedtke Professor of Marketing & Management Jones Graduate School of Management Rice University, Houston TX 77005 Contact: 713-348-6234, vmittal@rice.edu
- 2. n To start the process: » Determine how the research results will be implemented • Not enough to say we are looking for Segment X. If we identify segment X: – How will we communicate with them? Do we have data on their media using behavior? – How will we find members of segment X in a stream of new customers? Did we collect demographic data to allow us to identify them? » To ensure implementation of results, visualize results » Prepare dummy report » Backward Market Research: After you have agreement on the above two points, design the market survey to “fill in” dummy tables with real results. n Ensure that research is: » Relevant to current or anticipated decision » Timely » Efficient (a.k.a. cost effective / cheap) » Accurate » Usually can have any two, not all three n Manage the process » Be aware of the politics (document everything to avoid being the scapegoat) Ensuring Usable Research
- 3. Types of Marketing Research Studies n Cross-sectional study » Data collected at one point in time » Typically, survey based data » Typically, done on a custom basis and is descriptive » “Snapshot” n Longitudinal Study » From the same entity, data are collected at several points in time » Typically, behavioral data (e.g., credit-card purchases, monthly sales) » Typically, done on a syndicated basis; But trend toward custom » Descriptive and conclusive research goals » “Movie” n Syndicated versus Custom n Conclusive versus Descriptive
- 4. Types of Marketing Research Data n Attitudes and Perceptions n Collected via a survey n People provide “ratings” about specific topics (brand, lifestyle, issues, values) n Three components n Cognitive: What do you think about Supplier X? n Affective: How do you feel about Supplier X? n Behavioral: How do you intent to behave toward Supplier X? n Overall as well as attribute specific n What do you think of “billing accuracy” at Supplier X? n Overall, what do you think about Supplier X? n Intentions (to behave) n Likelihood to buy from Supplier X in the next six months n Intent to say positive things (word-of-mouth) to others about Supplier X n Intent to switch to a different supplier n Background and Classification n Demographics/Firmographics n Past behaviors (e.g., usage behavior, media behavior etc.) n Observational Data: NOT FROM SURVEY n Actual behaviors (e.g., actual sales from billing) n Customer purchases (based on loyalty card, checkout data) n Scanner data n Prescribing data on doctors
- 5. Perform ance M onitoring - Longitudinal KnowledgeRichness (Howwelldoweunderstandthephenomenon) Poor Managerial Purpose (What do we want to do with our results?) Understand (With more depth) Describe (With High Generalizability To Population of Interest) Manipulate (Make Causal Statements) Rich Exploratory - Qualitative - Ethnographic - Observational Descriptive - Survey Research - Secondary Data - Cross sectional Causal - Experimental - Longitudinal Different Types of Research: A Strategic View Understanding The “W hy” Exploring Blips, Deviations, & Patterns Strategic Advantage Through Differentiation
- 6. n Coverage Error: How comprehensive and representative is the sample list? n Also known as n Sampling Frame Error n Sample Bias/ Biased Sample n Sample Selection Bias n Under-selection (un-correctible) n Over-selection (easily correctible) n Non-representative sampling frame n Develop a comprehensive list n Exercise: What coverage errors can you detect in the healthcare survey? n Next step: Ensuring selection from sample list is unbiased, i.e., one type of respondent is not more likely to be selected All Research Is Wrong: Good Research is Useful Errors In Research Studies
- 7. n Sampling Error: To what extent are people selected from the sample list in an unbiased fashion? Stated differently, is selection from the list random? n Also known as n Margin of Error n Sampling Error n Larger sample size decreases sampling error n Ensure, that selection from that list is unbiased, i.e., one type of respondent is not more likely to be selected n VERY HARD TO DO!!!!!!! n Exercise: What would be the margin of error for the healthcare survey? Survey Errors All Research Is Wrong: Good Research is Useful
- 8. n Measurement Error: To what extent is the survey and its implementation done correctly? n Several Components n Survey fielding error (i.e., survey method error) n Survey design error (e.g., how survey questions are worded) n Interviewer error n Respondent error n Data coding, entry, and analysis error n For a given set of resources, larger sample decreases sampling error but increases measurement error n Exercise: What would be some potential survey errors in the healthcare study? Survey Errors All Research Is Wrong: Good Research is Useful
- 9. n Non response Error: Among those contact to participate, only a subset choose to participate? Among participants, not all questions are answered. n Key questions: How are responders different from non responders for the survey and for key questions? n Also evident in: n Low response rate error n High amount of missing data in surveys that were filled out n For a given set of resources, it is better to increase the response rate, than to have a larger sample size n Exercise: What would be some potential survey errors in the healthcare study? Survey Errors All Research Is Wrong: Good Research is Useful