Answer both questions:

1. A marketing research firm has conducted a ratings-based conjoint (ratings on 7-point scale)

study to measure consumer preferences for smartphones. They studied consumer preferences

for three smartphone attributes (Brand, Memory, and Price), with three levels for the brand and

memory attributes, and two levels for the price attribute:

Brand

Apple

Samsung

Google

Price

$600

128GB $800

256GB

Answer the following questions:

Memory

64GB

Here are the "partworths" (estimates) from this ratings-based conjoint:

Intercept

Apple

Google

128GB

256GB

$800

4.20

1.20

0.80

0.70

1.10

-1.00

1-1) The study adopted the base profile, "Google", "64GB", and "$600". What were the

"partworths" for these levels? Why is there a need for a base profile?

1-2) How many smartphone profiles did respondents evaluate if all possible profiles were

presented to them? What are some general strategies for overcoming a large number of profiles?

(Assume in this study the number of profile respondents evaluated was not a concern).

1-3) Interpret these estimates. How important is each the three attributes (Brand, Memory, and

Price)?/n2. Use ChatGPT to design and evaluate a conjoint study for a chatbot that will be used to assist

customer service representative in an Investment Bank. Ask ChatGPT to select 3 important

attributes and 2 levels of each attribute. These attributes/levels should naturally reflect trade-

offs that are common in new product designs. That is, the more of one desirable attribute can

make another less attractive for consumers. So, a chatbot that provides detailed responses may be

slower than one that provides less information, and so on.

Then ask Chat GPT to a) randomly sample potential financial service consumers. b) ask each

consumer to rate all (8) chatbot designs on a 100-point scale. Output the results in a CVS format

so that each row lists the consumer number, consumer rating, and chat bot attribute/levels.

2-1) Estimate the model: Copy and paste your results here:

2-2) Interpret these estimates. Do these results make sense to you? How important is each of the

three attributes?

2-3) Using your conjoint estimates, predict the rating for each chatbot. Predict the choice share

(probability) for each alternative using the share of preference rule.

2-4) How could you use these results for marketing research goals? How can you improve or

validate these findings?