Ralph's Rules of Presenting Empirical Social Science
Papers at Conferences
Intended Audience: Graduate Students
This short list is intended to complement Jerry Ratcliffe's Powerpoint presentation rules
You have two pieces to prepare. These are separate pieces. First, you have a talk. This is a script from which you will read. Second, you have a powerpoint presentation. These are different things that serve different purposes. The purpose of the script is so that you are not left stuttering inanely in front of dozens of people several of whom might have been -- until they heard your presentation -- future employers. The purpose of the Powerpoint is to give your audience some visual highlights linked to what they are hearing.
Preparing the Talk
If you are doing a typical empirical presentation, your paper will have four parts:
a literature review,
a description of your data sources and analysis plan,
a description of results,
and a reflection section offering a summary of what you have found, and what its implications are for theory policy and practice.
You want your presentation to be balanced, meaning that each of these four sections receives sufficient air time. Some typical mistakes made when constructing a paper for presenting are to either:
spend so much time on the literature review that you rush through the remaining portion of the paper at breakneck speed; or
spend so little time on the literature review that listeners have no idea what the important questions are in your area of research.
Literature. The literature review section should close with a clear statement of
the key hypotheses you will be testing, and
how those tests build upon or extend the previous work in an area.
Your listener wants to know both what is original about your effort, and how it ties to the past work. Key prior studies and authors should be mentioned by name, although you need not list all 16 authors of a paper.
If you are doing grounded theorizing, describe the kinds of connections or concepts you are hoping to find, and why and how they might be important.
You want to keep your key hypotheses section short: 2 or 3 at the most. Why? - because you are only going to have time to present 2 or 3 results (See below.)
Methods and analysis. In this section all your listener needs to know are the following:
What is the data source? If it is a secondary data source, give a brief outline of the original study. Tell folks how you came by the data. If it is primary data collection, tell the listeners generally how the data were gathered: who collected what information from whom using what kinds of instruments and when and where did this happen? Mind-numbing detail is not needed here. For example, if you have done a survey, you would want to talk briefly about the population, construction of the sampling frame, survey mode, types of items in the protocol, and response rates.
If your data structure is complicated by multiple waves or a panel design or some other longitudinal form, try and summarize graphically what is happening. For example, you could have a stacked bar chart showing total N and percent African-American by wave. You may need to spend some time thinking how best to summarize complicated data structures.
Tell listeners a little about the cases. If it is a survey, give gender, race and class backgrounds. If it is organizations or institutions, describe their sizes and purposes, and variation on key, theoretically-relevant predictors.
Specifically, what is the outcome variable or variables? If all else fails, be sure your readers can clearly see how your dependent variable was constructed, what its categories are, and what a higher vs. lower score means.
What are the predictors? Depending on how many predictors you have, you may or may not want to list all of them. But at least let listeners know what classes of predictors you have (demographics, attitudes, criminal career) and give them examples of predictors from each class.
What is your model or analysis plan? If you are doing something that can be communicated well with a graph, like an SEM or a causal analysis, then do so.
The key points here are:
not to get bogged down,
to be sure listeners get an idea of key data features, and
ensure your listeners know which aspects of the data are important to you, and how you will approach those data
Results. Just some, please!. Organize what you will cover, and compress as needed.
Focus only on the 2 or 3 most important specific results. Do not clobber listeners with a laundry list of findings. Decide, among the various "findings" you have, which are the most robust and theoretically relevant, and be sure that you clearly present those specific findings. (Or, if you are pain-seeking, select the most controversial results and present just those (tic)!) One of the biggest mistakes often made in this part of the presentation is to try and present a large number of results in serial form -- e.g. here are the results for the first outcome.... the second outcome.... the third outcome ..... ad nauseam. Listeners who are bored and stressed by too many bad papers they already have heard at the conference will tune you out pronto and start looking at the conference proceedings, fishing for the next session.
If you want to cover a lot of ground, compress the various results into one pattern, and present a summary table. For example, "we found across the three behavioral outcomes there was a significant impact of race, but for the attitudinal outcomes there was only one significant impact of race, and all results were in the theoretically expected direction." Once you have given people the one general pattern, provide details on one result just as an example, so they can get a "closer" look at the results.
Present just the basics of each result. You do not need to recite specific t-values or F values or chi squares or degrees of freedom or -2xLL values. The rule is: if you want to mention a specific result, do so only if you are going to help the listener interpret. For example: "The exponentiated coefficient for gender in the probit model was 1.30, suggesting that women, on average, as compared to men, on average, were 30% more likely to avoid a dangerous place in the neighborhood."
The result gaining the most attention will be the first one you present. Therefore, it should be the strongest one you have.
At the same time, stay honest. If your model worked for only one out of six outcomes, say that.
Discussion. Cover the following points:
Return to the questions or hypotheses you posed at the end of your introduction. What answers do you now have for these specific questions? Help the reader see how each answer ties to a specific result. Again, go over only the 2 or 3 most key questions.
Point out implications for theory. Again, keep the list short (no more than 2). If there are implications for policy or practice, mention a couple. If there are no policy implications, do not try and stretch it so it looks like there are.
You may want to briefly allude to limitations of your work.
Close with a sense of how this study may have pushed the knowledge frontier. What are the next questions?; How are the questions we are asking now different from those we were asking before?; What next steps will you be pursuing in your own work as you continue to investigate this topic?
Once you have your talk done, prepare your Powerpoint. It is a good idea to put breaks into your talk indicating where the overhead or slide changes.
GETTING READY FOR THE CONFERENCE
Practice giving your talk in front of an audience at least once. Offer to buy a pal a beer or a pizza or both. In addition, you want to practice reading through it several times, standing up, checking to see if you are reading your paper or delivering it.
You want to rehearse this several times. Why?
First, your talk is likely to be much too long, and is going to need cutting. So go through it once, time it, and then start cutting back. Present again to see if you are on target with the time. Although some session moderators at conferences are lackadaisical about time, many are not. You do not want to get cut off just as you were getting to the most important part.
Second, you are learning a script. When you get to the session you do not want to be reading from your script line by line. Rather you want to be delivering it, looking at the audience most of the time, and referring to the text when needed. You will be able to deliver it only if you are familiar with the text.
For your text, print it out in a way that is easy to read: I like 12 or 13 point Helvetica ALL CAPS, double or triple spaced. Be sure you have page numbers on the bottom.
Get Your Props Ready and Have Backups
If you are doing Powerpoint, have it on a couple of different disks. Floppies and cds go bad.
Be sure you have plain old overheads as a backup.
You also might want to have a brief handout. I usually like a 2 page handout with title, author, and abstract on the front, and a table or two of results on the back. Be warned, however, there is a danger with handouts that have tables in them: some person suffering from OCD (Obsessive Compulsive Disorder) in the back row is likely to ask you "How come the standard error in one equation is .07 and in another equation with the same predictor the standard error is .007, and doesn't that just invalidate all your results?" People pick at things they can hold and read. It is what we get trained to do.
If your session has a discussant, it is polite to send the paper to him/her at least a week before.
AT THE CONFERENCE
Find your session room in advance. Get a sense of the room, and how large an audience it will be able to hold.
On the day of your session, get to the room in advance. Find your moderator. Confirm how much time he/she will allow for your presentation. Find out what "time" cues he/she will give you. Find if questions will be after each paper or after all the papers.
Before you start your presentation, ask the audience to hold questions until the end.
During your presentation, have some way of keeping track of time.
Finally, there is a good chance that, despite all your preparation, you may be at a session with your paper placed last, a careless moderator, an audience that will not stay quiet, and voluble co-presenters, leaving you just 5 minutes to present your 20 minute paper. It happens; a lot. If you see this unfolding, think ahead to which parts of your Powerpoint are most important, and just jump to those. Rather than try and do the whole shebang, tell the audience your questions, and some of what you found. And if this happens, be gracious, not sour. It's just the way things are.