MRKT 600 Week 6 DB Forum 4 Example

Discussion Board Forum 4: Thread and Two Replies

MRKT 600 Applied Market Research, Discussion Board Forum 4 (thread and two replies)

Thread by [Student Name]

The prompt, the health system's directory test and the classmates' posts are composites standing in for the course's own forum; the textbooks and published studies are real; the requirements follow Liberty's public course guide for MRKT 600.

Thread

My health system is launching a new online directory of its primary care physicians next spring, and the marketing team is debating how each profile should look. Some colleagues want the star average at the top; others want a quoted review; others think a warm photograph and a sentence about the physician's approach to care will matter more. When we asked patients in a focus group, nearly everyone said the stars mattered most, but Moskowitz and Gofman (2007) argue that people often cannot tell us what actually drives their choices, and that systematically varying the elements of an offer and measuring responses reveals preferences better than asking.

Following their rule developing experimentation, I would test four elements, each with three options. The star summary could appear as a number, as stars only, or not at all. The review excerpt could be a specific comment about listening, a specific comment about access, or absent. The photograph could be a formal portrait, a casual photo in the exam room, or none. The care statement could describe the physician's approach, list credentials, or be absent. A planned set of nine combinations, an orthogonal design in which each option appears three times and every pair of elements is balanced, would keep the task short, and each respondent would rate every profile on a nine-point scale of how likely they would be to book. Alreck and Settle (2004) note that rating scales with clear anchors produce more consistent responses, so the scale's ends would be labeled definitely would not book and definitely would book.

Analysis would estimate how much each option adds to or subtracts from the likelihood rating, the same logic as the conjoint analysis Green and Srinivasan (1990) reviewed, in which the value of attributes is estimated from overall judgments of combined profiles. The limit is that respondents rate hypothetical profiles, and their ratings may not match what they do when booking a real appointment.

Classmates: what marketing decision in your work could be tested by varying elements rather than asking people what they prefer?

Reply to Kiara

Kiara, your experiment to test subject lines for your nonprofit's donation emails is a natural use of the method, and your four elements, urgency, a named beneficiary, a number and a question form, are the kinds of features Moskowitz and Gofman (2007) suggest combining. I wonder about one element, though. You listed "a number" with options of "a small number" and "a large number," which may be too vague to interpret; a subject line saying 12 families and one saying 1,200 families may differ in more than size.

Green and Srinivasan (1990) emphasize that attributes and their levels should be defined clearly and realistically so that respondents and analysts understand what each level means. You might replace that element with specific options, such as a number of people helped, a dollar amount, or no number, which would let you learn which kind of figure works best. Keeping the beneficiary element separate from the number element would also help, since a line that names one family and gives a dollar figure tests two ideas at once.

Alreck and Settle (2004) also caution about respondent fatigue. With four elements, could you limit each donor to eight subject lines so that the last ratings are as careful as the first?

Reply to Thomas

Thomas, your plan to test menu board designs for your coffee shop chain by having customers rate sixteen versions on a tablet is well organized and easy to run, and choosing elements the stores can actually change, item order, photographs, calorie display and price format, shows good judgment about what a test can actually change in the stores.

The one weakness I would point to is the one I noted in my own thread: ratings of hypothetical designs may not match what customers order. Green and Srinivasan (1990) discuss validation as an important issue for conjoint studies, and your stores give you a rare and inexpensive chance to check. After the experiment, you could install the two best-rated designs in two matched stores for a month and compare average ticket size and the share of featured items sold.

Moskowitz and Gofman (2007) present experimentation as a way to learn before committing, and a short field test would turn your ratings into evidence about behavior. The comparison would also show whether the design that customers rated highest actually raises sales or only looks appealing on a tablet. Would your regional manager agree to a one-month trial in two stores, with two similar stores kept as a baseline?

References

Alreck, P. L., & Settle, R. B. (2004). The survey research handbook (3rd ed.). McGraw-Hill/Irwin.

Green, P. E., & Srinivasan, V. (1990). Conjoint analysis in marketing: New developments with implications for research and practice. Journal of Marketing, 54(4), 3-19. https://doi.org/10.1177/002224299005400402

Moskowitz, H. R., & Gofman, A. (2007). Selling blue elephants: How to make great products that people want before they even know they want them. Wharton School Publishing.