Because that’s just good speaking. Chris Mooney has a very good post about the civic obligation scientists have to explain their work and its implications to the public. I recommend it highly. But one small part of the post rubbed me the wrong way (italics original; boldtype mine):
When it comes to scientific topics, citizens–and journalists, and policy-makers–want to know what the bottom line is, in plain language. They want to know why a topic matters, who it affects, what we can do about it. And can you blame them for feeling this way? There is a lot out there to pay attention to. We’re all suffering from information overload, all the time. It is very hard for anything to get through, much less anything technical or difficult.
This fact has huge implications for how scientists communicate, because it suggests an approach that runs strongly contrary to their instincts in many cases. Scientists are often prone to explain themselves through long, stepwise, technical arguments, eventually leading to some type of heavily hedged conclusion. So they’ll start out communicating like this: “I study X. X is a particular type of Y, found in Z. Previous researchers studying X had postulated that A and B most centrally influence its formation and development, but my work suggests that to the contrary, C plays the dominant role…”
And so forth. But what non-scientist is going to follow all the steps, trying to keep up with all the jargon and alien terms (here denoted by letters), without even knowing where it is all going to lead and why it matters?
That’s why scientists, in communicating, have to unlearn what they’ve learned in their training and put the conclusion first–followed by the details. For of course, once you understand why the details matter, you are more likely to grow interested in them and want to learn more. Yet this is very different from scientific instinct in many cases. It’s not how scientists are trained to talk to their peers.
Hunh? Because that’s not how I was trained to give a talk, nor is that a talk I would want to hear.
Even if you don’t put your all of your conclusions at the beginning of the talk, the main themes–and why you should spend an hour of your life listening to this–need to be upfront. Otherwise, it’s just a jumble of facts that don’t mean anything because they lack context. In fact, when I’ve been asked to help people review/practice their talks, that is the first thing that I (and usually everyone else in the room) beat out of the speaker.
For shorter talks, the twelve to twenty minute talks often given at conferences, I have a very simple formula, best described as ‘all of the Mad Biologist’s rhetorical gall can be divided into three parts’:
1) Part I: Tell them what you’re going to tell them.
2) Part II: Tell them what you’re telling them.
3) Part III: Tell them what you’ve just told them.
Hopefully, by the end, they might actually remember something (seriously, more than one or two key points in a 12 – 20 minute conference presentation is just a disaster waiting to happen).
I’m not saying that all scientists speak well (Sweet Baby Intelligent Designer that’s not true). But we’re not trained to speak poorly. Usually, poor speaking and keeping the punchline for the end is a reflection of no training at all.
Which is its own problem.
Nonetheless, Mooney still makes a lot of good points.
There’s nothing new here! Isaac Asimov described (in his autobiography?) how his thesis advisor criticized his first draft. Asimov’s write up began with all the evidence presented, then pieced it together and drew the conclusions out at the very end. Apparently it read more like a classic mystery novel, with all the suspects gathered in the drawing room in the last chapter, than as a scientific research report meant to communicate information.
I think Chris might be right. A few years ago, I guest-lectured on tech writing to a class of undergraduate electrical engineering students. When I told them they should begin their papers with their results, instead of saving the results for the conclusion, there was widespread rebellion. One student actually wrote a research paper for his Composition prof the next semester, explaining exactly why I was wrong. (That prof was my friend and told me–with amusement–all about the student’s paper as it progressed through the semester.) In short, I was ridiculed for my preposterous ideas. And yet, I also happen to teach business writing, a discipline that requires analysts to report their conclusions on the very first page, so their boss doesn’t have to wade through the entire thing unless they so desire. So, um, if electrical engineers want to work in actual businesses that might pay them, you know, actual salaries, they need to know that their business-educated bosses are going to require them to get to the point, toot suite.
It’s not just a matter of presentation, though. It’s a question of how you handle questions where the answer is somewhat uncertain. And here there is a real tension between the demands of science and policy. Scientific progress depends on a commitment not to affirm any claim that we don’t have rigorous evidence for, and to organize our work around attempts to falsify our preferred hypotheses. Whereas policy questions require us to take affirmative decisions even in the face of limited knowledge, and to advocate for those decisions. Obviously there are plenty of skills & habits that help with both, but fundamentally they are different modes of thinking and communicating.
Or as Peter Dorman puts it, the practice of science necessarily places a much greater weight on avoiding Type I than Type II errors. But
A practical problem with Chris’s suggestion when communicating with the public is that if you start off by presenting a point (in particular, a conclusion) that contradicts pre-existing conceptions, the audience begin looking harder for disconfirmation and for reasons to discount the information you are presenting. If you present the facts that lead to the conclusion first, while they may be less likely to follow your reasoning all the way through, they are more likely to at least get used to the steps along the way — steps that, presumably, will decrease the cognitive dissonance that must be overcome before acceptance of the final conclusion, in turn increasing the chance of eventual acceptance of the conclusion on future encounters with the line of reasoning.
Given that there is a large segment of society that is not merely ignorant of science, but that actively rejects many conclusions as dissonant to their worldview, Mooney’s suggestion seems naïve.
In support of my floccinaucinihilipilification of Mooney’s suggestion, I will note that the linked article presenting this argument does not follow its own advice. It builds up the reasoning, and then presents the conclusion. It would appear he has not even convinced himself.
As an aside, the three-part “tell ’em” approach is sometimes credited to Fermi, but Google Books indicates it dates back at least to 1911.
abb3w is right. Normally, and especially in technical writing, you start out with your conclusion. But when your audience is hostile to your conclusion, you start with something that they will agree with (or with a question). Otherwise, they won’t listen.
Most journals seem to put the context and conclusions up front in the abstract. I assume most people read this first. Why should a talk be structured so differently?