Steps to Improve the Reproducibility of Conclusions in the Social Sciences
Scientific conclusions often rely on generic statements (e.g., “X improves Y”), which may imply unwarranted generalisability. Prior research suggests that generic claims are perceived as more important than qualified statements (e.g., “X improves some of Y”), raising concerns about scientific communication. However, it remains unclear whether wording affects reproducibility—the extent to which different researchers reviewing the same data reach the same conclusion. To investigate this, I propose a vignette study where researchers evaluate study results and conclusions framed either generically or with a qualifier. If non-universal wording increases agreement, this suggests that qualifying claims improves robustness. This study builds on Multi100, a many-analyst project showing low analytical robustness in the social sciences. By exploring how language influences reproducibility, this research aims to improve best practices in scientific reporting and enhance the clarity and reliability of published findings.