Biostatistical Competencies and Attitudes in Health Research: Insights from a Delphi Study

Introduction: Although a large number of questionnaires on statistics self-efficacy (SE) and attitudes towards statistics (ATT) have been
developed in the literature, none have been specifically designed to investigate biostatistical literacy (BSL) among health researchers. However,
the importance of biostatistics is well established, as evidenced by its omnipresence in health-related training programs.
Objectives: 1) To determine fundamental statistical knowledge and skills that health researchers should have and that should be assessed in
terms of SE; 2) To draw up a list of attitudes that health researchers may have toward statistics.
Material and Methods: Forty-one items for SE and 27 items for ATT were first formulated and submitted to 25 panellists to obtain consensus
using a Delphi method. These items were developed from current literature such as SATS-36, Current Statistics Self-efficacy (CSSE) or the
statistical competencies in medical research listed by Oster and Enders (2018). The panellists were selected among biostatisticians,
psychometricians and health researchers from country's French-speaking universities using a snowball sampling. They were asked to rate each
item using a 7-Likert scale (strongly disagree to strongly agree). Consensus threshold was set at median ³ 6 and interquartile range (IQR) £ 1.
The total of round was set at 3, each lasting 3-4 weeks. Stability judgement was tested between the 1st and 2nd round using a Wilcoxon
signed-rank test.
Results: A total of 23 panellists (92%) took part in the 1st round. Consensus was reached on fundamental skills such as identifying a variable’s
measurement scale, distinguishing between population and sample parameters, and differentiating correlation from causality. The panellists also
emphasized that researchers must be able to understand, interpret, communicate and produce descriptive or basic inferential statistics, while
recognizing their limitations with complex analyses and consulting a biostatistician when needed. This item, not initially proposed, was widely
mentioned in the experts' comments and will be added in the 2nd round (November 2025). Finally, the experts also agreed that researchers
must acknowledge the usefulness of statistics in their training. During the 2nd round, 25 new items for SE and 41 for ATT were proposed to the
panellists based on their comments, in addition to the previous items. These included skills on position or dispersion parameters, or confidence
intervals. Attitudes related to effort and discouragement (to learn statistics) or overconfidence were also added. At the end of the 2nd round, the
same trends were observed for the skills, with judgment stability and a consensus among 21 (91%) experts on the added items. Same trends
also for attitudes, although a new consensus was reached on items related to difficulty (in learning) and effort. Finally, only items (11 for SE and
34 for ATT) that did not reach consensus or were modified and added based on comments from the 2nd round will be proposed to experts
during the 3rd round (January 2026).
Conclusion: A strong and consistent consensus emerged on essential biostatistical competencies and attitudes in health research. The final
round will address the remaining items to finalize the framework.