I believe the best results from a risk analysis meeting come from minimizing bias. The beginning of the meeting is about finding hazards by going through the chronological life cycle of the project. Starting with the machining/manufacturing process and following through till contact with end user. Once all the possible hazards are known, then break the meeting up into two groups. One assigns the hazards severity numbers and the other assigns frequency numbers. After both groups are finished, come back into one group to go over the numbers. Breaking into two smaller groups ensures that there is little bias in which hazards are the most risky. This makes it more believable when the group decides that some hazards are not risky enough to warrant a change.