- Rotterdam, Netherlands
Do you want to gain experience working for a marketing insights agency that influences the decisions of the world’s leading companies? We offer you the opportunity to write your thesis in collaboration with our research industry experts. In addition to writing your thesis, you’ll have the opportunity to step foot in the world of market research and work with an international team.
SKIM is a global insights agency helping leading companies thrive by understanding decision making. To stay ahead today, it’s critical to know how decisions are made and how the changing environment influences decisions for consumers, healthcare and B2B professionals. We combine decision-behavior know-how with analytical rigor, a thorough understanding of marketing challenges, and innovative research techniques. The result? Practical recommendations you can use to propel your business forward, online and offline.
Topic: Constraints in discrete choice estimation
Hierarchical Bayes routines have become the golden standard for estimating parameters in discrete choice experiments in which every participant evaluates a series of scenarios.
The most commonly used commercial HB software for this application is CBC-HB, developed by Sawtooth Software Inc.
A feature that Sawtooth introduced in their software is the allowance of constraints. For example, users of the software may constrain price utilities to become negative, or generally to constrain all ordinal attributes with a natural preference order. The advantages of these constraints are:
- It reduces noise during the estimation
- It will never lead to ‘counter-intuitive’ predictions
Sawtooth provides two variants of prohibitions. One is a very simple post hoc correction. The other is a more advanced routine that incorporates prohibitions on the fly in a specific manner. An obvious disadvantage of the former solution is that the outcomes are no longer model-based. A big disadvantage of the latter solution is that the parameters of the upper model converge to different ranges depending on the random starting point.
For this internship we have two big questions:
- Which other implementations of constraints are there currently (literature search) or can we come up with ourselves (idea generation)?
- Which implementation of constraints works better, while looking at hit rates, MAE and convergence? This would be based on the analysis of real data
You have solid analytical skills, you are good company and you have a good sense of humor. Preferably you are/have:
- In the final stage of your Master, preferably in Business Administration, Econometrics, Economics, Marketing, Statistics, Life Sciences or Psychology, with above-average grades
- You have a conceptual mind and are proficient in modeling
- Strong analytical skills
- Proficiency in R
- Excellent written and oral communication skills in English
- Proactive attitude expressed in extracurricular activities
- Keen interest in quantitative and/or qualitative market research methodologies
- Ability to quickly adapt to a complex, dynamic environment
- Authorization to work in the Netherlands
With this internship we want to give you the opportunity to seriously explore what a career at SKIM will be like. In case expectations from both ends are met during the internship period, you might be recruited for a starter position at SKIM.
We offer this assignment as an internship for a Master’s degree. We align our conditions with those of your school. We offer a fun and friendly place to work, dedicated guidance, a multi-cultural team, and decent financial compensation. Our office is located right next to Rotterdam Central Station.
Please submit your applications, consisting of a resume and cover letter, written in English. Due to the large volume of applications, only applicants who are invited to the first round selection process will be further contacted (this may take 1-3 weeks). Thank you for your understanding.