*Please note this opportunity is only available to UK and EU applicants*
Procter & Gamble is one of the largest consumer goods companies worldwide, specialising in a range of personal care, healthcare, and homecare products. P&G perform many iterations of consumer research to drive product innovation. One area of research uses mathematics, statistics and data science to link consumer satisfaction with technically measured product performance vectors that are known to be important in that category. That is, we aim is to create predictive models that relate the affective response or attitudes of the Consumers to the Performance of products (C2P modelling). These models are used as a key innovation tool to guide the design / formulation of products enabling R&D to innovate faster and at lower cost while still allowing us to meet the changing needs of our consumers.
This research project is concerned with exploring the application of modern measurement theory in consumer-to-performance (C2P) modelling for evaluation and development of consumables. The aim is to apply modern measurement theory, and in particular, the Rasch model, to C2P modelling. Modern measurement theory is proposed as it establishes a linear measurement of consumers’ affective response and/or attitudes to the product and allows us to examine the responses of individuals. We can generate calibrated distributions on both the likelihood of a population to endorse category products and the endorsability of the difference statements assessed in the questionnaire. In the context of consumer research, it can be used to assess the quality of consumer research data to determine the validity of using the data in subsequent C2P modelling and we hope to demonstrate usefulness in building predictive models.
The research aims to establish protocols, standards and tools for the implementation of modern measurement to consumer goods categories and can be separated into 8 individual research questions;
1. Assessment of software packages available to conduct Rasch analyses (Rumm, Winsteps, R) and implement analysis process using the software that is most applicable to structure of consumer survey data plus demonstrates optimal reliability and analytical metrics .
2. Propose standards for fit statistics in the context of consumer research data to allow assessment of models and their performance.
3. Construct calibrated questionnaire scales from which we will be able to predict product overall rating by analysing existing consumer research data (e.g. ordinal question response data), demonstrate reliability across several data applications.
4. Link the use of Rasch measurements to predict overall acceptance / attitudes back to measurements of physical properties of the products (C2P modelling) and compare with use of conventional approaches.
5. Using 3 & 4 demonstrate the ability to:
i. Test small samples with future prototypes and predict overall product rating for a larger population
ii. From technical measures use C2P modelling and predict the overall rating of a new prototype product comparing results with field data to assess accuracy.
6. Establish guidelines for optimal design of consumer research field studies to enable Rasch measurement.
7. Explore and demonstrate any benefit of computerized adaptive testing in the implementation of calibrated questionnaires for consumer research and compare with use of computerized adaptive testing in traditional research methods.
8. Demonstrate the use of calibrated questionnaires, implemented via CAT to assess the reliability of testing small samples that represent a new user group (segment or country) and compare results with existing groups to identify proxies for future research.
Project ref: LE44
Application deadline: 1st June (midnight)