• Select a problem or question of interest in your work, in your daily life or in some area that is of interest to you such as politics, family life or a sporting activity.
• Define the question in such a way that the answer (and any decision that might be made as a result) depends on a single parameter.
• Decide what data you could collect to help estimate the parameter and hence make the decision. Note that the data may be as limited as a single value or fictitious but sensible ones.
• Choose and justify the distribution from which you think the data will come.
• Now consider and justify the form of the prior distribution for this parameter. Using the method suggested by O'Hagan, or otherwise, completely specify the prior distribution.
• Hence derive the posterior distribution
• Choose and justify a sensible cost function (you may use more than one if you want). Hence find the estimate of the parameter of interest, so answering the question, solving the problem and/or making the relevant decision.
• If you can, set up a credible interval for the parameter.
• Produce a short well-written report of the question and its solution. Technical details can be put into clearly referenced, concise appendices if this seems appropriate.