M21 - Tutorial Conjoint Analysis

Question 1

Conjoint analysis

–> what is your prference order for the products?

Answer 1

• method used to quantify the utility perceived by a single person
• -> individual analysis
• objects are seen as a bundled set of attribute values
• considered jointly

Question 2

What are the two types of basic classes of methods for conjoint analysis?

Answer 2

Traditional

Choice-based

Question 3

Traditionalconjoint analysis

Answer 3

• objects are ranke din order of prefernce
• based on this ordinal ranking, metric part-worth utilities fo the individual attribute values of the objects are estimated
• -> sum of. Partworth utilities = total utility of objects
• evaluator has a totally deterministic preference model

Question 4

Choice-based conjoint model

Answer 4

• evaluator makes selection decision put of the considered set of objects
• no ranking but selection of the best alternative ( no choice is also possible)
• estimation of metricpartworth utilities for the individual attribute values of the objects based on ordinal ranking
  Sum of utilities partworths = total utility per object

Question 5

Adv of conjoint analysis

Answer 5

• valid in evaluating price - how much a customer is willing to pay for a good
• variables can be categorial, rather than continuous

Question 6

Conjoint procedure

Answer 6

1 attributes and attribute valued
2 survey design 
2a definition of stimuli
2b number of stimuli
3 evaluation of stimuli
4 estimation of the utility values
5 aggregation of the utility values

Question 7

Attributes and attribute values

Attributes must be...
1
2
3
Attribute values must be ...
4
5
Attributes and attribute values must be...
6
7

Answer 7

Attributes must be…
1. relevant for pref decision
2 manipulable in reality
3 should be independent

Attribute values…
4 feasible
5 in a compensatory relship with each other

Attributes and attribute values must be..
6 not constitute criteria for exclusion
7 must be few

Question 8

Stimulus

Def

Methods

Answer 8

= combination of attribute values

1. profile method: one value of each attribute
2. two-factor method: only two attributes are included and compaired pairwise

Question 9

Survey design - number of stimuli

The number of stimuli … … with the number of … …

• 2 designs

Answer 9

Survey design - number of stimuli

The number of stimuli INCREASES EXPONENTIALLY with the number of ATTRIBUTE VALUES

• full design - usually not practical
• reduced design - subwst of stimuli that represengs tje full design as well as possible
• -> random sample
• -> if all attributes have the same no of values –> symmetric design asymmetric

Symmetric design = latin design

Question 10

Assessment of the stimuli methods

Answer 10

Rankings, scale ranks, pairwise comparison

Question 11

Estimating the utility values

How?

Answer 11

Based on the ranking data, partworth utilities are calculated for all attribute values by means of conjoint analysis

Partworth values by regr analysis
–> assumption: ranking values are equidistand –> metric

Question 12

Aggregation of utility values

Only possible after what?

2 possibilities to generate aggregated results:

What to do when data is very heterogeneous?

Answer 12

• comparison and aggregation of individual analysis is possible only after normalizing the calvulated part-worth utilities
• -> common zero point and common scaling
• 2 possibilities:
  1. individual analysis and aggregation of obtained partworth utilities
  2. common conjoint an. for several persons delivering aggregated partworth utilities
• perform cluster analysis first