Conjoint Analysis and MaxDiff Webinar Q&A


We would like to again thank everyone who signed up for our webinar that aired on Wednesday, December 4th, 2013 at 1PM Eastern titled "How to Use Conjoint and MaxDiff to Put Your Best Product Forward." Close to 600 marketers and market research professionals signed up and nearly 200 joined us for the live broadcast. There were many questions from our audience that we did not have the opportunity to answer in the end when we ran out of time.

Survey Analytics VP of Client Services Esther LaVielle took a moment yesterday afternoon to answer questions we did not have the opportunity to get to. Thanks again to everyone that joined us! If you missed out on this presentation, you can access the replay here.

Q1: What all checks are available in Data Validation? 

Esther: In terms of data validation for both the Conjoint and Maxdiff after a study has been concluded you will be able to access both an online report as well as all of the raw data. You will be able to see what’s been displayed and what was picked and draw your own conclusion if needed.

If you are speaking of validation requirements prior to deploying a survey you can access our survey overview page which highlights any errors in logic skip patterns prior to deployment. The best thing I would recommend is to run a test among your team and also colleagues who are not familiar with conjoint to solicit feedback on the user experience to give you time to clearly define any features or attributes you are using for either question type. All test data can be deleted prior to deployment and will not influence your final results.

Q2: Which is more beneficial - Conjoint or Maxdiff? 

Esther: My personal opinion is that all questions, even conjoint and maxdiff, are beneficial so long as people can clearly understand and answer it promptly. Each question type is unique in it’s data output so it would depend on what your end goal of your project is. Are you looking to figure out what new products or services to sell at certain in 2014? Then Choice based conjoint would be a good fit. Maxdiff, on the other hand, can be applied to many areas of research including brand preference, customer satisfaction, product or service features, and even message or image testing. The good thing about Maxdiff is that the list of items you want to present to the respondent does not.

Q3: Can we create conjoint in Survey Analytics and call it in mail survey in confirmit? 

Esther: We offer survey URL links and where you post them is on your own discretion, however, we do not integrate with competitor software. You should be able to have the capability to run a comprehensive survey directly in our platform if required.

Q4: Hello - I missed the beginning of this, I'm sorry. WIll this be more about the application of Conjoint and max diff, or about how to set them up technically speaking?

Esther: Yes, towards the end of the presentation recording at about 40 min in, you will see a demonstration of how to set up both question types in our system. If you would like to run an evaluation on our site, please feel free to sign up for an evaluation license here: https://surveyanalytics.com/a/showEntry.do?mode=surveyanalytics.

Q5: Output from the simulators is preference share, not truly market share. Does the model make assumptions (or does user have a choice) to get from preference share to a truer estimate of actual market share? 

Esther: We use the term market share as an estimator based on the data that’s been collected. In this case the term preference share can be interchangeable as you are simulating the idea of showing particular profiles to the respondents and based on what they’ve answered we have the output. If you are speaking about the true definition of market share then using an estimation tool based on the data received will not achieve this. Definite spot-on accuracy of market share is not possible. If anything it will give you an idea of potentiality within the market space of the product/services in consideration. •

Q6: When should you use MaxDiff v Discrete Choice? 

Esther: You should use conjoint analysis if you are looking to gather information on potential market share of new or existing products/services you are looking to make money on. It’s been used as part of a pricing research strategy as well as test new ideas prior to determine if investment and resources should be poured into it. Maxdiff can be used a part of many different types of research to determine preferences among your attribute list. You can use it in customer satisfaction studies, brand awareness, product or service, or message/image testing. Anytime you are faced with big rating scale you should ask yourself if possibly using a maxdiff question type would give you insights that you may potentially miss with standard questions.

Q7: What is the pricing model.. Is it an annual license? 

Esther: We offer monthly and annual licenses here at Survey Analytics. Please feel free to reach out to our sales team to get you started with an evaluation license or discuss pricing. [email protected] Phone: 1-800-326-5570

Q8: Does your platform integrate with Confirmit and Qualtrics?

Esther: We do not integrate with competitor software. You can easily create and deploy a comprehensive survey to be deployed across all channels within Survey Analytics. We have a free trial available if you would like to further explore the tool and what it offers: https://surveyanalytics.com/a/showEntry.do?mode=surveyanalytics.

Q9: How do you define "homogenous" for d-optimize? 

Esther: We use the Merriam-Webster definition. http://www.merriam-webster.com/dictionary/homogeneous

Q10: Can you do a market share simulation with MaxDiff?

Esther: For conjoint, yes we can, but for Maxdiff that is not typically done. However it’s a great idea and something I can speak to my head developer and operations manager on. If you would like to send us more information as to why you think this would be valuable to add please contact me directly at [email protected]. It would be great to speak to you about it.

Q11: What is your experience in terms of the % of people who are invited that will actually participate?

Esther: Well, in any research project we know that about 50% of your budget is allocated to recruitment, sample finding, accessing and managing a database of willing participants, and incentives. The better you are at managing the respondents who fully understand what is asked of them, and the more clear and concise the survey is, then the higher the response rates. Other things like incentives, timing of survey deployment, survey topic, follow up reminders, and etc. will affect your response rates as well.

If you are interested in conjoint and maxdiff in Survey Analytics specifically I will give you a few tips that can help:

  1. Make sure your features and levels are clear and easy to understand and in the language of your target respondents. For maxdiff, make sure the attributes are easy to understand and if you need, limit the number shown per page.
  2. Set up definitions, photos, instructions, and tips within the conjoint or maxdiff or prior to getting to the conjoint so they know what is expected of them. 
  3. Too man conjoint exercises will result in respondent fatigue and will skew your end results. Make sure to keep it as concise as possible. If you are using Doptimal design and it says to run a larger task count, make sure you are clear with the respondent how many you questions you must run by them, and let them know the incentive is directly correlated to their responses for these questions. 
  4. Target consumers who WANT to take this kind of survey. Pre-screen if necessary. 
  5. Use all follow up tools to remind those who have not completed to take the survey. Survey Analytics offers a reminder scheduler that you can set up ahead of time with a message targeting those who need to complete the survey.

Comments