Photo by Letizia Bordoni on Unsplash
Recently, I have worked with the president of Singapore Actuarial Society in performing conjoint analysis.
Hence, in this post, I will summarize what I have learnt about conjoint analysis. There will be no demonstration in this post.
What is “conjoint analysis”?
Conjoint analysis is a form of statistical analysis that firms use in market research to understand how customers value different components or features of their products or services. It’s based on the principle that any product can be broken down into a set of attributes that ultimately impact users’ perceived value of an item or service (Stobierski 2020).
(Heiss 2023) also explained that conjoint experiments are a special kind of randomized experiment where study participants are asked questions that have experimental manipulations. However, unlike a standard randomized experiment where one feature of interest is manipulated (like in an A/B test), conjoint experiments are choose-your-own-adventure randomized experiments.
Different types of conjoint analysis
(Stobierski 2020) listed the different types of conjoint analysis:
| Conjoint Analysis | Description |
|---|---|
| Choice-Based Conjoint (CBC) Analysis | This is one of the most common forms of conjoint analysis and is used to identify how a respondent values combinations of features. |
| Adaptive Conjoint Analysis (ACA) | This form of analysis customizes each respondent’s survey experience based on their answers to early questions. It’s often leveraged in studies where several features or attributes are being evaluated to streamline the process and extract the most valuable insights from each respondent. |
| Full-Profile Conjoint Analysis | This form of analysis presents the respondent with a series of full product descriptions and asks them to select the one they’d be most inclined to buy. |
| MaxDiff Conjoint Analysis | This form of analysis presents multiple options to the respondent, which they’re asked to organize on a scale of “best” to “worst” (or “most likely to buy” to “least likely to buy”). |
Areas conjoint analysis can be applied
Conjoint analysis can be applied in the following areas (Stobierski 2020):
Pricing
Sales & marketing
Research & development
Some of the important concepts under conjoint analysis
Variable importance
Similar to how it works in a machine learning context, this measurement will show which variable is more important in explaining the result.
Part-worth/utility values
This would show the amount of weight an attribute level carries with a respondent (Singh 2023).
Minimum number of responses to be considered as credible
To be able to conclude the results are credible, below is the formula to compute how many responses we need from the conjoint analysis (Qualtrics):
\[Minimum\ number\ of\ response\ =\frac{Constant\times Maximum\ number\ of\ levels\ in\ any\ feature}{Number\ of\ choices\ per\ question\times\ Number\ of\ question}\]
Where the constant is 750 if the total number of levels across all features is less than or equal to 10, and 1,000 if the total number of levels across all features is greater than 10.
Below are what we could do if we felt the minimum number of responses required is too high:
Increase the number of questions
Increase of the choices per question
Decrease the maximum number of levels in the feature
Helpful materials
Following are some of the helpful materials I found online:
How to interpret utility score
Conclusion
That’s all for the day!
Thanks for reading the post until the end.
Feel free to contact me through email or LinkedIn if you have any suggestions on future topics to share.
Refer to this link for the blog disclaimer.
Till next time, happy learning!
Photo by Gregoire Jeanneau on Unsplash