Toward Transparent Model Selection in Marketing Research: An Open-Science Decision Protocol for Documenting, Justifying, and Reporting Modelling Choices

Abstract

Open science has secured the openness of marketing research's inputs, data, and code, but not the accountability of the reasoning behind the model a study reports. This paper names that gap the transparency paradox of model selection. It locates the layer that pre-registration, code sharing, multiverse and specification-curve analyses, and recent selection criteria for PLS-SEM each leave opaque, then develops the Model-Selection Transparency Protocol (MSTP): a methods-agnostic, five-stage decision-documentation spine with reusable artefacts, demonstrated on genuine European Customer Satisfaction Index data. Two defensible specifications fit comparably, yet four selection criteria do not converge decisively on either, and the choice between them implies opposite managerial conclusions — precisely the situation in which documented reasoning, not a further statistic, is what the field needs. The MSTP gives researchers, reviewers, and editors a usable means of documenting, justifying, and reporting modelling choices across statistical and interpretive traditions, reframing transparency from the sharing of inputs to the accountability of reasoning.