The panel comprised biostatisticians, health researchers and psychometricians who were affiliated with, or collaborated with, French-speaking universities in Belgium. Researchers developed a study-specific questionnaire from the literature and existing questionnaires, then asked panellists to rate each item on a seven-point Likert scale and provide qualitative feedback. Items could be revised, added or removed between rounds. Consensus required both a median score of at least 6 and an interquartile range no greater than 1. The researchers assessed response stability between the first two rounds using the Wilcoxon signed-rank test.

The broad agreement on core competencies indicates that the panel viewed a shared baseline of statistical ability as important across different kinds of health research. The strongest consensus concerned fundamental biostatistical knowledge, interpreting statistical information, communicating statistical findings and critically appraising statistical claims. By contrast, specialised statistical methods and regression models were associated with lower agreement. The authors suggested that this pattern may reflect variation among researcher profiles: techniques considered necessary for one role may not be expected to the same extent in another. The study does not provide a universal ranking of advanced methods, but instead indicates where the panel did not agree on common requirements.

Agreement was substantially lower for statements about attitudes towards biostatistics. Panellists were comparatively more consistent in endorsing the perceived value of biostatistics and expressing interest in the subject. Responses concerning anxiety and discouragement showed greater variability, and differences appeared between expert groups, particularly regarding the expected degree of autonomy in biostatistical practice and perceptions of statistical complexity. These results show that some attitude-related propositions were more readily accepted than others; they do not establish how common positive engagement or discouraging experiences were among health researchers as a whole.

The Delphi design produces a structured account of expert consensus rather than a direct measurement of researchers’ competence or attitudes. It therefore does not show that health researchers possess the agreed skills, nor does it test whether education based on the proposed framework improves research quality, learning outcomes or clinical practice. The panel was limited to experts connected with French-speaking Belgian universities, so the framework should not automatically be treated as applicable to other countries, disciplines or institutional settings. The authors explicitly call for further validation before wider use. Within the study setting, however, the findings offer a context-specific basis for developing assessment tools, training programmes, curricula and lifelong learning strategies, and for evaluating researchers to identify training needs.

Subject of Research: Expert consensus on core biostatistical competencies and attitudes among health researchers in French-speaking Belgium

Article Title: Core biostatistical competencies and attitudes towards biostatistics among health researchers: a Delphi study

Article References: Dardenne, N., Sayed, S. E., Diep, A., Bruyere, O., & Donneau, A.-F. (2026). Core biostatistical competencies and attitudes towards biostatistics among health researchers: a Delphi study. BMC Medical Education. https://doi.org/10.1186/s12909-026-10208-8

Image Credits: AI Generated

DOI: 10.1186/s12909-026-10208-8

Keywords: biostatistics, health research, Delphi study, research competencies, expert consensus, statistical literacy, critical appraisal, research training, Belgian universities, Core, biostatistical, competencies