Scenario #3: Government Research
A large number of statisticians are employed in the public sector. The largest employer of statisticians in Canada is Statistics Canada, and statisticians also work in many government research institutes and laboratories including those run by the Ministries of the Environment and Agriculture. Part of a statistician’s role in government positions is to relate statistical concepts to civil servants and politicians who will use this information to guide their policy decisions. Imagine the scenario:
After receiving your PhD studying the asymptotic behaviour of likelihood methods applied to various estimators for the number of planets in our solar system, you have landed a job at the Department of Agriculture with a research group studying the effects of spontaneous moose grazing on Alberta's production of alfalfa – the province’s number one export. A year ago you began by designing some experiments to compare the effects of different anti-moose agents on alfalfa production. These experiments were conducted over the summer and the data has been analysed under your supervision. You have now been asked to present your results to the government task force studying moose control. You will have 1 hour to present your results.
Summarize your ideas for giving a successful presentation in this situation. Here are some questions to help guide your thinking, although they are by no means exhaustive:
1. What are the objectives of your presentation? How will this guide the content of your presentation?
It is important to identify our audience and realize that they are not statisticians so that we should stay away from statistical jargon as much as possible. Also restrict ourselves to discussing how the study addresses the question of interest, but leave off what excites us as statisticians!!
We have a concrete question, and we need to provide a concrete answer.
Justify your funding!!! Show how hard you worked with lots of emphasis on how data was collected (using helicopters, hi-tech equipment, etc., make it sound like CSI)
Considering that there is a value judgement in the question of interest, we can provide specific steps for maximizing alfalfa production.
This means that our content will not be specific to statistical methods, but to how the results of the experiment can be used to maximize alfalfa production.
2. What aspects of your results would you emphasize for the policy makers? How would you suggest they incorporate uncertainty into their actions?
The politician just wants the point estimate and our interpretation of it. Talk about confidence intervals (with a brief explanation) in terms of how it provides information into the certainty in our results. Small intervals=high certainty (we need to protect our own reputations with at least a few caveats!!!)
Caution about limitations to and sampling design, and the risks of extrapolating beyond the actual experimental data.
3. Suppose you found that one moose-deterrent which outperformed all others. What evidence would you present to the task force and how would you describe it?
I would show them a boxplot (or something similar) which had the best moose deterrent in bright pink so that it stands out against all other options. If there are actual picture from the field (nice field vs. bad field) I would show that picture. Compare also the return on investment that we get by using this fantastic method.
4. What are some general presentation dos and don’ts? Can you think of any specific dos and don’ts for the scenario described?
Considering to whom we are presenting(politicians), 0 mathematics and lots of graphs and pictures.
Tables clear, concise, and no more than a few columns
Wear a nice suit and shoes!! Otherwise people might not take you seriously.
Firm handshake and remember to make eye contact.
Speak slowly and clearly(imagine you are having a conversation with George Bush)
Slides should have no more than one main point, and no more than 3 bullets
Avoid all technical details and nuance that you may think is important. Ignore issues like whether it is an observational study, and focus on conclusions: are there more moose now than 5 years ago, what is your recommendation based on this
Do time it so that you finish early, and leave lots of room for questions eg. 15 minutes per hour
Use hidden slides so that you can answer questions about technical details in case you do get questions
Use lots of idiomatic language that politicians will appreciate (not statistical!) eg. ‘Push the envelope’, ‘look to the future’