Sunday, February 27, 2011

Seminars by the Number

If I were not aware of the post hoc ergo propter hoc fallacy, I’d say it was entirely my fault. Last weekend I cleaned off the barbeque and re-filled the propane tank. It’s a good way to spend a mild and sunny February afternoon in Victoria, because I don’t have to take responsibility for cutting the grass since we down-sized to a townhouse. Inevitably, we’ve just had a week with an unusually heavy snowfall, icy roads and (by our standards) chilly temperatures. These conditions, coupled with the fact that it was the Spring “reading break” at UVic, meant that the corridors of the “grove academe” were rather quiet. In particular, the usual research seminars for faculty and graduate students were not scheduled.

Over the years, I’ve participated in more economics seminars than I care to think about. Some of them were memorable. Some of them were memorable for the wrong reasons. Some of them brought out the worst in me, and are better forgotten. Very few of the seminars that are held in our department are really “econometrics seminars”, but that’s O.K.. There’s more to life than just having fun, after all. One thing that most of our graduate students don’t realize, however, is that the style and tone of seminars in (any branch of) economics are quite different from those of seminars in many other disciplines. Take seminars in statistics, for example. The following observations are based on my regular attendance at such seminars in multiple institutions over the past 35 years or so:


                                                       Economics                              Statistics

Duration                                          1.5 hours                                 50 minutes
Explanation of details                        Heavy                                      Minimal
Focus on “the big picture”               Occasional                               Almost always
General tone                                  Often confrontational                 Always polite
Questions at end of seminar              Several                                    One or two
Audience interruptions                      Continuous                               None


Of course, I realise that the above comparisons are somewhat over-simplified. It’s traditional for economics seminars to be prefaced by remarks from the chair to the effect that the audience may wish to raise ‘points of clarification’ during the presentation, but that time should be left at the end of the seminar for ‘more substantive’ questions and discussion. Traditionally, the speaker nods in understanding, and says how they welcome any questions at any time. The history of this tradition lies, by the way, with the econometricians at the Cowles Commission. Not only has this piece of history largely been forgotten, but the ‘Cowles rules’ are largely ignored in practice. Typically, this results in seminars in which there are constant interruptions from the audience – sometimes on the pretext of seeking ‘clarification’ – often making it difficult for the speaker and audience alike to focus on the real content of the presentation. Now, some will say that this all makes for a lively seminar. However, as economists we also know that there are opportunity costs associated with seminars that follow this pattern.

In the spirit of improving this situation, I have a modest proposal that I’d like to put forward. Let me confess from the outset that a suggestion exactly along these lines was made many years ago by Ross Williams (now Emeritus Professor of Economics at the University of Melbourne in Australia). It’s a suggestion that is worth recalling, and if nothing else, it’s in need of an update. I can’t remember if Ross put his idea in writing, or if it was just something that came up at dinner (after an especially trying seminar or conference presentation). Being an econometrician, Ross couched his suggestion in terms best suited to that sub-discipline, but the basic idea can be extended, almost trivially (as we like to say) to any economics seminar or conference session.

In the interests of streamlining an econometrics seminar presentation, all members of the audience would use an agreed-upon numeric code when asking questions. There is a fairly standard menu of questions that arise in seminars - the hardy perennials that go with the territory – and these could be codified. For example, the inevitable question ‘Did you test your data for unit roots’ could be assigned the number 4. Of course, even more efficiency could be introduced by encouraging the speaker to use a standard numeric code as well. On this side of the table matters would be quite simple: 1 = ‘Yes’; 2 = ‘No’; 3 = ‘That’s a good question’; 4 = ‘I tried it, but it didn’t affect the results’; and 5 = ‘It seemed like a good idea at the time’. Keen-eyed readers will notice that there is no code for, ‘I don’t know’.

A suitable code for audience members requires a little more thought. When this idea catches on (as it surely will) we could consider having a specific coding system for time-series seminars; a slightly different one for seminars dealing with cross-section data issues; and so on. The generic speaker’s code suggested above would suffice in all cases. For the meantime, though, here is a suggested basic coding to get things going:


Code Number            Question/Comment (Disruption)

1                                  Why didn’t you use a Logit/Probit model instead of OLS?
2                                  Did you test for weak instruments?
3                                  Are those standard errors ‘heteroskedasticity-consistent’?
4                                  Did you test your data for unit roots?
5                                  Are your data seasonally adjusted, or not?
                                    (This one is tricky – see the dialogue below.)
6                                  Couldn’t you simply bootstrap the confidence intervals?
7                                  Do you think that the number of replications that you used
                                    in your Monte Carlo experiment is really enough?
8                                  What about structural breaks in your
                                    data?                           
9                                  This would be much more easily accomplished if you took
                                    a Bayesian approach.
10                                I don’t understand why you are ignoring the obvious
                                    pre-test issues that arise here.
11                                Does this test have good power in the present context?
12                                Why are you trying to estimate a model that is obviously
                                    under-identified?
                                    (One of my personal favourites.)


As a refinement, and to avoid disturbing those members of the audience who really do know about opportunity cost, and have fallen asleep mid-way through the seminar, a verbal exchange of numbers could be replaced by having participants silently raise little ‘paddles’ with the desired numbers on them. I have in mind the sort of thing that is used at public auctions to identify bidders.  

To illustrate the obvious merits of the proposed system, here is a real-life example of an exchange of views at an econometrics seminar;

Dr. A:                          1?       
Speaker:                      4.
Dr. A:                          3?
Speaker:                      1.
Dr. Z:                           8?
Speaker:                      3. Yes,......, 3.
Dr. McQ:                     5?
Speaker:                      2 (and therefore 1).
Drs. D, A, E:                11, 6, 6           
Speaker:                      1, 3, 4.
Dr. A:                           But 12?
Speaker:                       5.

etc.

Now there's a lively seminar!







© 2011, David E. Giles

7 comments:

  1. Awesome. Can we please try this out at the next family dinner?

    ReplyDelete
  2. Excellent! David, I have a place to learn from you again. Don't forget I am looking forward to working with you always. :) -----Qian

    ReplyDelete
  3. hahaha, that is brilliant.

    the next step, which is obviously impossible, is for all those question askers to just 'take a number' so that the speaker can just get through the material.

    ReplyDelete
  4. John - I wish I had thought of that!!!

    ReplyDelete
  5. This book Dave? http://www.amazon.com/Generalized-Edition-Monographs-Statistics-Probability/dp/0412317605/ref=sr_1_1?ie=UTF8&qid=1335973781&sr=8-1

    ReplyDelete
    Replies
    1. Thomas - thanks - but I think I missed your point.

      Delete