Data collection: Treat every variable as a treasure

• The difficulties of quantitative analysis of collected data are under-estimated.
• The human factor is the principal source of bias.
• Consensus between observers participating in data collection is essential.
• Data collection software should be developed step-by-step, guided by quantitative and qualitative analysis after each step.

Collection of data concerning case histories is not yet common in homeopathy despite its great importance for this method. Computer program development progresses slowly and discussion about requirements is scarce.Two Dutch projects assessed Materia Medica of some homeopathic medicines and six homeopathic symptoms. Especially the second project relied heavily on data collection. In both projects much effort was spent on consensus between participating doctors.There was much variance between doctors despite our consensus efforts. Assessing causality seems the most important source of bias, there is also much variance in assessing symptoms.ConclusionData collection software should be developed step-by-step, guided by close monitoring and feedback of participating practitioners.