Lecture 4 Flashcards

1
Q

Conjoint Analysis

  • Purpose
  • Can Help To
A

One of the most used market research methods

Purpose: Estimate the importance of a specific product feature (Wi)

Can Help To:

  • Decide whether to incorporate/improve a specific feature
  • Estimate products’ market shares
  • Estimate product demand
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Rating-Based Conjoint Analysis

A

Idea: Ask respondents to rate their valuation (V) of products with different levels of attributes (Ai)
-Uses linear regression to estimate the weight of each attribute (Wi)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Rating-Based Conjoint Analysis Pros and Cons

A

Pros:
a. Corresponding to consumer choice theory
b. Straightforward to implement
Cons:
a. Hard for respondents to rate alternatives in a consistent way
-Solution: choice-based conjoint analysis
b. Too many alternatives (when # attributes increases)
-Solution: adaptive design for choice-based conjoint
c. Do not have incentive to think carefully
-Solution: incentive-alignment for choice-based conjoint

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Choice-Based Conjoint Analysis

A

Each respondent will be asked a set of the questions

  • choosing from 2-4 product alternatives, with different combination of attributes
  • answers help us estimate the weight of each attribute
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Importances Report

A

Importance value of screen:

                         Wscreen ---------------------------------------------------------- Wscreen + Wstorage + WfaceID + Wprice
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

More Choice-Based Conjoint Analysis

A

Decide whether to incorporate/improve a specific feature
-Consumer’s willingness-to-pay for an improvement in attribute i:
(THIS IS CALLED “PARTWORTH”)

WTPi: Wi (*) change in price level
—————————-
Wp

When an attribute changes to something else, how much more or less is a consumer willing to pay

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Estimating Product Market Share

A
  • Different consumers can have different values of Wi (they assign different weight to each attribute) –> They have different favorite product
  • Demand share of a product = proportion of consumers that have it as the most preferred product
  • In practice, the software estimates the distribution of Wi, and then calculate proportion of consumers who choose the product = the product’s market share
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Product Market Share (Vs.) Product Demand

A

Market Share does not equal Product Demand

Product Market Share:
 # of consumers who buy A / # of consumers who buy a product in the category (that A belongs to)
Product Demand: 
# consumers who buy A/ (consumers who buy in the category + consumers who do not buy)

**To estimate product demand, we need to know how many consumers’ won’t buy (None Option) **

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Incentive-Aligned Choice-Based Conjoint

A

Respondents’ choices will have a consequence, so that respondents are motivated to choose carefully.

Results from incentive-aligned choice-based conjoint have better prediction power that hypothetical ones.

Implementation: Refer to Slides

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Adaptive Choice-Based Conjoint

A

When the number of attributes increases, there will be huge number of combinations (a combination = a possible product)

Respondents have limited attention so we can optimally choose the product sets in each question to make the estimation more efficient & more accurate
-Choosing the next, question adaptively
(Questions on Sawtooth are chosen adaptively)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly