|
Elements of the
Conjoint Data Collection and Analysis Process
A brief review of
the stages of the conjoint analysis data collection process and the
analysis elements are useful in creating the attributes and their
associated level (feature descriptors) to be assessed in the
research. In creating the attribute set it is important to think
ahead with respect to the outcomes that can be expected and to make sure
that the concepts evaluated in the research can are meaningful to the
respondents and translatable into actionable properties that marketers
can use to produce and differentiate a product.
Stages of
the Data Collection Process:
Three primary data
collection stages are involved in the approach CME (Adaptive Conjoint
Analysis Model) will utilize to accomplish the research objectives of
this project:
Stage 1
- Initial Preference Ranking of Remaining Acceptable Attribute
Levels.
The first stage of
the process the respondent is asked to rank order the levels within
attributes which do not have an obvious ranking. With the
elimination of obvious ranked levels (those exhibiting a
logical/rational monotonic order) the respondent is then asked to
indicate his first, second, etc. choice of levels (features) for each
attribute:
Example Question: Indicate which brands
would be your first choice, second choice, etc assuming all other
considerations are equal:
A)
Honda
B) Toyota
C) GM
Stage 2 – Importance
Rating of Attribute Levels
The next stage, the
respondent is asked to define the degree of relative importance of each
attribute level defined as acceptable based on the preference
differences. The objective of this stage is to isolate and
calibrate those product features having the strongest emotional
attachment. It may be noted that initial utility estimates are generated
during this phase of the research.
Example Question:
If two vehicles were both acceptable in all other ways, how important
would this difference be if one was Honda the other Toyota?
A)
Honda
Versus
B) Toyota
Stage 3
– Trade-Off (Paired Concept) Analysis Stage
It is at this stage
that final utility levels (parts-worth’s) are calibrated using a
regression technique. The respondent is placed in a conflict
situation at this point and forced to trade-off (paired concept) having
relevant feature combinations:
Example Question: Note which combination
would be most preferable to you:
A)
Honda
Costing $25,000
Lifetime Warranty Replacement
Or
B)
Toyota
Costing $20,000
7 year Warranty Replacement
In the above
case the respondent is being asked to trade-off brand against cost and
the type of warranty. The paired comparison section may give
respondents up to 5 feature combinations. Generally we prefer to
use 2 or 3 features at a time.
Stage 4 – Final
Analytical Stages:
Based on these
three primary data collection stages, along with a final calibration and
consistency (validity) check, the final utility values are generated for
each respondent. These individual utility value profiles may be
used in additive models to define the respondent’s optimal and
sub-optimal product features. Segmentation analysis of these utility
sets will show how groups of respondents view these aggregate product
profiles. For example, Mean differences in the utility levels of
an attribute define the relative importance of each attribute and its
features to the segment. What if analyses (simulations) can also
be performed to show the relative impact of substituting one feature for
another with respect to first choice models, purchase likelihood models,
etc.
Simulation Analysis
Phase:
In addition,
specific analyses by CME will provide simulation findings for the key
brand price interactions. It is suggested that the simulation
analyses use a likelihood of purchase simulation measure to define the
various levels of the price attributes relative to brand
attributes. The primary research issues “driving” the simulation
analysis will focus on isolating comparative differences and identifying
the potential manufacturers that would be potential targets for premium
pricing.
|
