I've been doing a bunch of presentations recently to both policy and academic audiences and been reflecting a bit on what the differences in presenting to these two different kinds of audience. Here are a couple of thoughts -- additional contributions are welcome as this is probably a topic that could take up a blog of its own.
1. Keep the language universal. If you want to reach the whole audience, you have to keep the language at a level that everyone can understand. This is pretty obvious, but there are a couple of traps here.
Trap #1: the hidden geek. You are presenting to a policy audience and a bespectacled nerd asks about your standard errors. With glee you launch into a discussion of the relevant clustering and robustness. The problem I have faced is that this nerd has now unleashed my inner nerd, and it's hard to put that genie back in the bottle. But it's critical to switch back -- speak briefly to the technical question in the appropriate language and then get back to being comprehensible to all.
Trap #2: dumbing it down. One mistake I have made is to get too simplistic. One of the beautiful things about identification is that this an idea that doesn't take economics training to get it. And more than once I have caught myself making things too simplistic and getting all caught up in a discussion that could have been handled much better with more sophistication at the outset.
Trap #3: avoid the jargon, all of it. This is another slippery slope. It's so easy to say linear relationship and so much harder to say it a more accessible way. The great thing about technical language is that's precise and brief. But in some cases, a higher quantity of accessible words will be better (kind of like translating things into French).
2. Be prepared for out-of-the-box questions. A room full of policy folks will usually have multiple disciplinary backgrounds, and there is a high chance that a sociologist or a scientist or even an accountant will come at you from an angle you were not prepared for.
3. There is also a high chance there is someone in the audience who knows the context better than you do. Yes, we're development economists, and yes, we spend a lot of time in the field. But my experience with policy audiences is that they'll either know different things about my local context or something about the national or international context that's relevant and I don't know. I've definitely been humbled by this more than once.
4. Think about what's interesting (hint: it's not your triple difference or the fact that you've come up with the perfect instrument). Methodology is secondary. Innovations in measurement are tertiary (unless they are totally counterintuitive). It's the policy conclusions and why they are important that are key. Again this is perhaps obvious, but my experience is that it takes a lot of self control to not talk about how cool the survey was.
5. A debunking result is not enough. This was a lesson I learned from one of my (policy) mentors at the World Bank. His point is that knocking down established wisdoms is a great way to get people's attention, but then the key is offering them something constructive in their place.
6. Think about the practical implications for your results. Odds are high there is someone in the audience is going to think about actually how one might implement your policy recommendations, so spend some time thinking beforehand how this might actually work in practice. It's one thing to say the results imply the need for better targeting. It can be quite another to actually do that targeting.
7. Be prepared to talk about external validity. This is a question I see a lot and I think the right approach is to say what you can about your context, and try to draw what external validity there might be without throwing up your hands and saying this is only valid for 4 villages in southern Uganda.
8. In this vein, also be prepared to talk about cost effectiveness. If you are convincing policy makers that what your results are showing is a good idea for them to adopt, they're going to want to know what it'll cost them.
9. Don't use tables in your presentation. At least not the ones with standard errors, asterisks, or balance tests. I often use the language of statistically significant, but don't put this in table form.
10. Convey your enthusiasm. My sense in academic seminars is that enthusiasm can be seen as detracting from some notion of scientific impartiality. Not so for a policy audience.
11. Don't be afraid to ask for clarification. I often get questions I don't understand. In some academic seminars, asking people to explain can either signal my ignorance or suggest weakness. In policy seminars, my experience is that the diversity of people can lead to questions coming from places I hadn't thought of. And they usually don't mind explaining.
12. Keep it shorter.
1. Keep the language universal. If you want to reach the whole audience, you have to keep the language at a level that everyone can understand. This is pretty obvious, but there are a couple of traps here.
Trap #1: the hidden geek. You are presenting to a policy audience and a bespectacled nerd asks about your standard errors. With glee you launch into a discussion of the relevant clustering and robustness. The problem I have faced is that this nerd has now unleashed my inner nerd, and it's hard to put that genie back in the bottle. But it's critical to switch back -- speak briefly to the technical question in the appropriate language and then get back to being comprehensible to all.
Trap #2: dumbing it down. One mistake I have made is to get too simplistic. One of the beautiful things about identification is that this an idea that doesn't take economics training to get it. And more than once I have caught myself making things too simplistic and getting all caught up in a discussion that could have been handled much better with more sophistication at the outset.
Trap #3: avoid the jargon, all of it. This is another slippery slope. It's so easy to say linear relationship and so much harder to say it a more accessible way. The great thing about technical language is that's precise and brief. But in some cases, a higher quantity of accessible words will be better (kind of like translating things into French).
2. Be prepared for out-of-the-box questions. A room full of policy folks will usually have multiple disciplinary backgrounds, and there is a high chance that a sociologist or a scientist or even an accountant will come at you from an angle you were not prepared for.
3. There is also a high chance there is someone in the audience who knows the context better than you do. Yes, we're development economists, and yes, we spend a lot of time in the field. But my experience with policy audiences is that they'll either know different things about my local context or something about the national or international context that's relevant and I don't know. I've definitely been humbled by this more than once.
4. Think about what's interesting (hint: it's not your triple difference or the fact that you've come up with the perfect instrument). Methodology is secondary. Innovations in measurement are tertiary (unless they are totally counterintuitive). It's the policy conclusions and why they are important that are key. Again this is perhaps obvious, but my experience is that it takes a lot of self control to not talk about how cool the survey was.
5. A debunking result is not enough. This was a lesson I learned from one of my (policy) mentors at the World Bank. His point is that knocking down established wisdoms is a great way to get people's attention, but then the key is offering them something constructive in their place.
6. Think about the practical implications for your results. Odds are high there is someone in the audience is going to think about actually how one might implement your policy recommendations, so spend some time thinking beforehand how this might actually work in practice. It's one thing to say the results imply the need for better targeting. It can be quite another to actually do that targeting.
7. Be prepared to talk about external validity. This is a question I see a lot and I think the right approach is to say what you can about your context, and try to draw what external validity there might be without throwing up your hands and saying this is only valid for 4 villages in southern Uganda.
8. In this vein, also be prepared to talk about cost effectiveness. If you are convincing policy makers that what your results are showing is a good idea for them to adopt, they're going to want to know what it'll cost them.
9. Don't use tables in your presentation. At least not the ones with standard errors, asterisks, or balance tests. I often use the language of statistically significant, but don't put this in table form.
10. Convey your enthusiasm. My sense in academic seminars is that enthusiasm can be seen as detracting from some notion of scientific impartiality. Not so for a policy audience.
11. Don't be afraid to ask for clarification. I often get questions I don't understand. In some academic seminars, asking people to explain can either signal my ignorance or suggest weakness. In policy seminars, my experience is that the diversity of people can lead to questions coming from places I hadn't thought of. And they usually don't mind explaining.
12. Keep it shorter.
Join the Conversation